Ciencia y tecnología

We live surrounded by AI-generated summaries. We have had the option of generating them for months, but now they are imposed on digital platforms as the first content that our eyes see when using a search engine or opening an email thread. On platforms such as Microsoft Teams or Google Meet, video call meetings are transcribed and summarized in automatic minutes for those who have not been able to be present, but also for those who have been there. However, what a language model has considered important, is it really important for the person receiving the summary?

In this new context, the key is to learn to recover the meaning behind so much summarized information. These three strategies will help you transform automatic content into an understanding and decision-making tool.

1. Ask expansive questions

We tend to summarize to reduce content that we are not able to cover, but we run the risk of associating brief with significant, an equivalence that is not always fulfilled. Therefore, we should not focus from the beginning on summarizing, but on extracting relevant information for us, our context, our vision of the situation and our way of thinking. Beyond the  basic prompt "give me a summary", this new way of approaching content that escapes us consists of cross-referencing data, connecting dots and suggesting hypotheses, which they call  sensemaking. And it happens, first of all, to be clear about what we want to know.

Practical situation:

Imagine a long meeting that we have not been able to attend. That afternoon, we received in our email a summary of the topics discussed. It's not always possible, but a good practice at this point, if our organization allows it, is not to just stay with the summary: if allowed, and always respecting confidentiality guidelines, upload the full transcript to a conversational system such as Copilot or Gemini and ask specific questions:

• Which topic was repeated the most or received the most attention during the meeting?

• In a previous meeting, person X used this argument. Was it used again? Did anyone discuss it? Was it considered valid?

• What premises, assumptions or beliefs are behind this decision that has been made?

• At the end of the meeting, what elements seem most critical to the success of the project?

• What signs anticipate possible delays or blockages? Which ones have to do with or could affect my team?

Beware of:

First of all, review and confirm the attributions. Generative models are becoming more and more accurate, but they have a great ability to mix real information with false or generated information. For example, they can attribute a phrase to someone who did not say it, relate ideas as cause and effect that were not really connected, and surely most importantly: assign tasks or responsibilities for next steps to someone who does not correspond.

2. Ask for structured content

Good summaries are not shorter, but more organized, and the written text is not the only format we can use. Look for efficiency and ask conversational systems to return tables, categories, decision lists or relationship maps. Form conditions thought: if you structure information well, you will understand it better and also transmit it better to others, and therefore you will go further with it.

Practical situation:

In this case, let's imagine that we received a long report on the progress of several internal projects of our company. The document has many pages with paragraphs descriptive of status, feedback, dates, unforeseen events, risks and budgets. Reading everything line by line would be impossible and we would not retain the information. The good practice here is to ask for a transformation of the document that is really useful to us. If possible, upload the report to the conversational system and request structured content in a demanding way and without skimping on details:

• Organize the report in a table with the following columns: project, responsible, delivery date, status, and a final column that indicates if any unforeseen event has occurred or any risk has materialized. If all goes well, print in that column "CORRECT".

• Generate a visual calendar with deliverables, their due dates, and assignees, starting on October 1, 2025 and ending on January 31, 2026, in the form of a Gantt chart.

• I want a list that only includes the name of the projects, their start date, and their due date. Sort by delivery date, closest first.

• From the  customer feedback section  that you will find in each project, create a table with the most repeated comments and which areas or teams they usually refer to. Place them in order, from the most repeated to the least.

• Give me the billing of the projects that are at risk of not meeting deadlines, indicate the price of each one and the total.

Beware of:

The illusion of veracity and completeness that a clean, orderly, automatic text with fonts will provide us is enormous. A clear format, such as a table, list, or map, can give a false sense of accuracy. If the source data is incomplete or wrong, the structure only makes up the error and we will have a harder time seeing it. AI productions are usually almost perfect. At the very least, and if the document is very long, do random checks ignoring the form and focusing on the content.

3. Connect the dots

Strategic sense is rarely in an isolated text, let alone in a summary. The advanced level in this case consists of asking the multimodal chat to cross-reference sources, compare versions or detect patterns between various materials or formats, such as the transcript of a meeting, an internal report and a scientific article. What is really interesting to see are comparative keys such as evolutionary changes, absences or inconsistencies.

Practical situation:

Let's imagine that we are preparing a proposal for a new project. We have several materials: the transcript of a management team meeting, the previous year's internal report, and a recent article on industry trends. Instead of summarizing them separately, you can upload them to the same conversation thread or chat you've customized on the topic, and ask for more ambitious actions.

• Compare these three documents and tell me which priorities coincide in all of them, even if they are expressed in different ways.

• What topics in the internal report were not mentioned at the meeting? Generate a hypothesis for each one as to why they have not been treated.

• What ideas in the article might reinforce or challenge ours? Give me ideas that are not reflected in our internal report.

• Look for articles in the press from the last six months that support the strong ideas of the internal report.

• Find external sources that complement the information missing in these three documents on topic X, and generate a panoramic report with references.

Beware of:

It is very common for AI systems  to deceptively simplify complex discussions, not because they have a hidden purpose but because they have always been rewarded for simplicity and clarity in training. In addition, automatic generation introduces a risk of authority: because the text is presented with the appearance of precision and neutrality, we assume that it is valid and useful. And if that wasn't enough, structured summaries are copied and shared quickly. Before forwarding, make sure that the content is validated, especially if it contains sensitive decisions, names, or data.

AI-based models can help you visualize convergences, gaps, or contradictions and, from there, formulate hypotheses or lines of action. It is about finding with greater agility what is so valuable that we call insights. That is the step from summary to analysis: the most important thing is not to compress the information, but to select it well, relate it and connect it with the context. Intensifying the demand from the prompt is the most appropriate way to work with AI systems, but it also requires a previous personal effort of analysis and landing.

Content created by Carmen Torrijos, expert in AI applied to language and communication. The content and views expressed in this publication are the sole responsibility of the author.

Data visualization is a fundamental practice to democratize access to public information. However, creating effective graphics goes far beyond choosing attractive colors or using the latest technological tools. As Alberto Cairo, an expert in data visualization and professor at the academy of the European Open Data Portal (data.europa.eu), points out, "every design decision must be deliberate: inevitably subjective, but never arbitrary." Through a series of three webinars that you can watch again here, the expert offered innovative tips to be at the forefront of data visualization.

When working with data visualization, especially in the context of public information, it is crucial to debunk some myths ingrained in our professional culture. Phrases like "data speaks for itself," "a picture is worth a thousand words," or "show, don't count" sound good, but they hide an uncomfortable truth: charts don't always communicate automatically.

The reality is more complex. A design professional may want to communicate something specific, but readers may interpret something completely different. How can you bridge the gap between intent and perception in data visualization? In this post, we offer some keys to the training series.

A structured framework for designing with purpose

Rather than following rigid "rules" or applying predefined templates, the course proposes a framework of thinking based on five interrelated components:

1. Content: the nature, origin, and limitations of the data
2. People: The audience we are targeting
3. Intention: The Purposes We Define
4. Constraints: The Constraints We Face
5. Results: how the graph is received

This holistic approach forces us to constantly ask ourselves: what do our readers really need to know? For example, when communicating information about hurricane or health emergency risks, is it more important to show exact trajectories or communicate potential impacts? The correct answer depends on the context and, above all, on the information needs of citizens.

The danger of over-aggregation

Even without losing sight of the purpose, it is important not to fall into adding too much information or presenting only averages. Imagine, for example, a dataset on citizen security at the national level: an average may hide the fact that most localities are very safe, while a few with extremely high rates distort the national indicator.

As Claus O. Wilke explains in his book "Fundamentals of Data Visualization," this practice can hide crucial patterns, outliers, and paradoxes that are precisely the most relevant to decision-making. To avoid this risk, the training proposes to visualize a graph as a system of layers that we must carefully build from the base:

1. Encoding

1. It's the foundation of everything: how we translate data into visual attributes. Research in visual perception shows us that not all "visual channels" are equally effective. The hierarchy would be:

• Most effective: position, length and height
• Moderately effective: angle, area and slope
• Less effective: color, saturation, and shape

How do we put this into practice? For example, for accurate comparisons, a bar chart will almost always be a better choice than a pie chart. However, as nuanced in the training materials, "effective" does not always mean "appropriate". A pie chart can be perfect when we want to express the idea of a "whole and its parts", even if accurate comparisons are more difficult.

2. Arrangement 

1. The positioning, ordering, and grouping of elements profoundly affects perception. Do we want the reader to compare between categories within a group, or between groups? The answer will determine whether we organize our visualization with grouped or stacked bars, with multiple panels, or in a single integrated view.

3. Scaffolding

Titles, introductions, annotations, scales and legends are fundamental. In datos.gob.es we've seen how interactive visualizations can condense complex information, but without proper scaffolding, interactivity can confuse rather than clarify.

The value of a correct scale

One of the most delicate – and often most manipulable – technical aspects of a visualization is the choice of scale. A simple modification in the Y-axis can completely change the reader's interpretation: a mild trend may seem like a sudden crisis, or sustained growth may go unnoticed.

As mentioned in the second webinar in the series, scales are not a minor detail: they are a narrative component. Deciding where an axis begins, what intervals are used, or how time periods are represented involves making choices that directly affect one's perception of reality. For example, if an employment graph starts the Y-axis at 90% instead of 0%, the decline may seem dramatic, even if it's actually minimal.

Therefore, scales must be honest with the data. Being "honest" doesn't mean giving up on design decisions, but  rather clearly showing what decisions were made and why. If there is a valid reason for starting the Y-axis at a non-zero value, it should be explicitly explained in the graph or in its footnote. Transparency must prevail over drama.

Visual integrity not only protects the reader from misleading interpretations, but also reinforces the credibility of the communicator. In the field of public data, this honesty is not optional: it is an ethical commitment to the truth and to citizen trust.

Accessibility: Visualize for everyone

On the other hand, one of the aspects often forgotten is accessibility. About 8% of men and 0.5% of women have some form of color blindness. Tools like Color Oracle allow you to simulate what our visualizations look like for people with different types of color perception impairments.

In addition, the webinar mentioned the Chartability project, a methodology to evaluate the accessibility of data visualizations. In the Spanish public sector, where web accessibility is a legal requirement, this is not optional: it is a democratic obligation. Under this premise, the Spanish Federation of Municipalities and Provinces published a Data Visualization Guide for Local Entities. 

Visual Storytelling: When Data Tells Stories

Once the technical issues have been resolved, we can address the narrative aspect that is increasingly important to communicate correctly. In this sense, the course proposes a simple but powerful method:

1. Write a long sentence that summarizes the points you want to communicate.
2. Break that phrase down into components, taking advantage of natural pauses.
3. Transform those components into sections of your infographic.

This narrative approach is especially effective for projects like the ones we found in data.europa.eu, where visualizations are combined with contextual explanations to communicate the value of high-value datasets or in datos.gob.es's data science and visualization exercises.

The future of data visualization also includes more creative and user-centric approaches. Projects that incorporate personalized elements, that allow readers to place themselves at the center of information, or that use narrative techniques to generate empathy, are redefining what we understand by "data communication".

Alternative forms of "data sensification" are even emerging: physicalization (creating three-dimensional objects with data) and sonification (translating data into sound) open up new possibilities for making information more tangible and accessible. The Spanish company Tangible Data, which we echo in datos.gob.es because it reuses open datasets, is proof of this.

Figure 1. Examples of data sensification. Source: https://data.europa.eu/sites/default/files/course/webinar-data-visualisation-episode-3-slides.pdf

By way of conclusion, we can emphasize that integrity in design is not a luxury: it is an ethical requirement. Every graph we publish on official platforms influences how citizens perceive reality and make decisions. That is why mastering technical tools such as libraries and visualization APIs, which are discussed in other articles on the portal, is so relevant.

The next time you create a visualization with open data, don't just ask yourself "what tool do I use?" or "Which graphic looks best?". Ask yourself: what does my audience really need to know? Does this visualization respect data integrity? Is it accessible to everyone? The answers to these questions are what transform a beautiful graphic into a truly effective communication tool.

Artificial Intelligence (AI) is becoming one of the main drivers of increased productivity and innovation in both the public and private sectors, becoming increasingly relevant in tasks ranging from the creation of content in any format (text, audio, video) to the optimization of complex processes through Artificial Intelligence agents.

However, advanced AI models, and in particular large language models, require massive amounts of data for training, optimization, and evaluation. This dependence generates a paradox: at the same time as AI demands more and higher quality data, the growing concern for privacy and confidentiality (General Data Protection Regulation or GDPR), new data access and use rules (Data Act), and quality and governance requirements for high-risk systems (AI Regulation), as well as the inherent scarcity of data in sensitive domains limit access to actual data.

In this context, synthetic data can be an enabling mechanism to achieve new advances, reconciling innovation and privacy protection. On the one hand, they allow AI to be nurtured without exposing sensitive information, and when combined with quality open data, they expand access to domains where real data is scarce or heavily regulated.

What is synthetic data and how is it generated?

Simply put, synthetic data can be defined as artificially fabricated information that mimics the characteristics and distributions of real data. The main function of this technology is to reproduce the statistical characteristics, structure and patterns of the underlying real data. In the domain of official statistics, there are cases such as the United States Census , which publishes partially or totally synthetic products such as OnTheMap (mobility of workers between place of residence and workplace) or SIPP Synthetic Beta (socioeconomic microdata linked to taxes and social security).

The generation of synthetic data is currently a field still in development that is supported by various methodologies. Approaches can range from rule-based methods or statistical modeling (simulations, Bayesian, causal networks), which mimic predefined distributions and relationships, to advanced deep learning techniques. Among the most outstanding architectures we find:

• Generative Adversarial Networks (GANs): a generative model, trained on real data, learns to mimic its characteristics, while a discriminator tries to distinguish between real and synthetic data. Through this iterative process, the generator improves its ability to produce artificial data that is statistically indistinguishable from the originals. Once trained, the algorithm can create new artificial records that are statistically similar to the original sample, but completely new and secure.

• Variational Selfencoders (VAE): These models are based on neural networks that learn a probabilistic distribution in a latent space of the input data. Once trained, the model uses this distribution to obtain new synthetic observations by sampling and decoding the latent vectors. VAEs are often considered a more stable and easier option to train compared to GANs for tabular data generation.

• Autoregressive/hierarchical models and domain simulators: used, for example, in electronic medical record data, which capture temporal and hierarchical dependencies. Hierarchical models structure the problem by levels, first sampling higher-level variables and then lower-level variables conditioned to the previous ones. Domain simulators code process rules and calibrate them with real data, providing control and interpretability and ensuring compliance with business rules.

You can learn more about synthetic data and how it's created in this infographic:

 

Figure 1. Infographic on synthetic data. Source: Authors' elaboration - datos.gob.es.

While synthetic generation inherently reduces the risk of personal data disclosure, it does not eliminate it entirely. Synthetic does not automatically mean anonymous because, if the generators are trained inappropriately, traces of the real set can leak out and be vulnerable to membership inference attacks. Hence, it is necessary to use Privacy Enhancing Technologies (PET) such as differential privacy and to carry out specific risk assessments. The European Data Protection Supervisor (EDPS) has also underlined the need to carry out a privacy assurance assessment before synthetic data can be shared, ensuring that the result does not allow re-identifiable personal data to be obtained.

 Differential Privacy (PD) is one of the main technologies in this domain. Its mechanism is to add controlled noise to the training process or to the data itself, mathematically ensuring that the presence or absence of any individual in the original dataset does not significantly alter the final result of the generation. The use of secure methods, such as Stochastic Gradient Descent with Differential Privacy (DP-SGD), ensures that the samples generated do not compromise the privacy of users who contributed their data to the sensitive set.

What is the role of open data?

Obviously, synthetic data does not appear out of nowhere, it needs real high-quality data as a seed and, in addition, it requires good validation practices. For this reason, open data or data that cannot be opened for privacy-related reasons is, on the one hand, an excellent raw material for learning real-world patterns and, on the other, an independent reference to verify that the synthetic resembles reality without exposing people or companies.

As  a seed of learning, quality open data, such as high-value datasets, with complete metadata, clear definitions and standardized schemas, provide coverage, granularity and timeliness. Where certain sets cannot be made public for privacy reasons, they can be used internally with appropriate safeguards to produce synthetic data that could be released. In health, for example, there are open generators such as Synthea, which produce fictitious medical records without the restrictions on the use of real data.

On the other hand, compared to a synthetic set, open data allows it to act as a verification standard, to contrast distributions, correlations and business rules, as well as to evaluate the usefulness in real tasks (prediction, classification) without resorting to sensitive information. In this sense, there are already works, such as that of the Welsh Government with health data, which have experimented with different indicators. These include total distance of change (TVD), propensity score and performance in machine learning tasks.

How is synthetic data evaluated?

The evaluation of synthetic datasets is articulated through three dimensions that, by their nature, imply a commitment:

• Fidelity: Measures how close the synthetic data is to replicating the statistical properties, correlations, and structure of the original data.

• Utility: Measures the performance of the synthetic dataset in subsequent machine learning tasks, such as prediction or classification.

• Privacy: measures how effectively synthetic data hides sensitive information and the risk that the subjects of the original data can be re-identified.

Figure 2. Three dimensions to evaluate synthetic data. Source: Authors' elaboration - datos.gob.es.

The governance challenge is that  it is not possible to optimize all three dimensions simultaneously. For example, increasing the level of privacy (by injecting more noise through differential privacy) can inevitably reduce statistical fidelity and, consequently, usefulness for certain tasks. The choice of which dimension to prioritize (maximum utility for statistical research or maximum privacy) becomes a strategic decision that must be transparent and specific to each use case.

Synthetic open data?

The combination of open data and synthetic data can already be considered more than just an idea, as there are real cases that demonstrate its usefulness in accelerating innovation and, at the same time, protecting privacy. In addition to the aforementioned OnTheMap or SIPP Synthetic Beta in the United States, we also find examples in Europe and the rest of the world. For example, the  European Commission's Joint Research Centre (JRC) has analysed the role of  AI Generated Synthetic Data in Policy Applications, highlighting its ability to shorten the life cycle of public policies by reducing the burden of accessing sensitive data and enabling more agile exploration and testing phases. He has also documented applications of multipurpose synthetic populations for mobility, energy, or health analysis, reinforcing the idea that synthetic data act as a cross-sectional enabler.

In the UK, the Office for National Statistics (ONS) conducted a Synthetic Data Pilot to understand the demand for synthetic data. The pilot explored the production of high-quality synthetic microdata generation tools for specific user requirements.

Also in health , advances are observed that illustrate the value of synthetic open data for responsible innovation. The Department of Health of the Western Australian region has promoted a Synthetic Data Innovation Project and sectoral hackathons where realistic synthetic sets are released that allow internal and external teams to test algorithms and services without access to identifiable clinical information, fostering collaboration and accelerating the transition from prototypes to real use cases.

In short, synthetic data offers a promising, although not sufficiently explored, avenue for the development of artificial intelligence applications, as it contributes to the balance between fostering innovation and protecting privacy.

Synthetic data is not a substitute for open data, but rather enhances each other. In particular, they represent an opportunity for public administrations to expand their open data offering  with synthetic versions of sensitive sets for education or research, and to make it easier for companies and independent developers to experiment with regulation and generate greater economic and social value.

Content created by Jose Luis Marín, Senior Consultant in Data, Strategy, Innovation & Digitalisation. The content and views expressed in this publication are the sole responsibility of the author.

Education has the power to transform lives. Recognized as a fundamental right by the international community, it is a key pillar for human and social development. However, according to UNESCO data, 272 million children and young people still do not have access to school, 70% of countries spend less than 4% of their GDP on education, and 69 million more teachers are still needed to achieve universal primary and secondary education by 2030. In the face of this global challenge, open educational resources and open access initiatives are presented as decisive tools to strengthen education systems, reduce inequalities and move towards inclusive, equitable and quality education.

Open educational resources (OER) offer three main  benefits: they harness the potential of digital technologies to solve common educational challenges; they act as catalysts for pedagogical and social innovation by transforming the relationship between teachers, students and knowledge; and they contribute to improving equitable access to high-quality educational materials.

What are Open Educational Resources (OER)

According to UNESCO, open educational resources are "learning, teaching, and research materials in any format and support that exist in the public domain or are under copyright and were released under an open license." The concept, coined at the forum held in Paris in 2002, has as its fundamental characteristic that these resources allow "their access at no cost, their reuse, reorientation, adaptation and redistribution by third parties".

OER encompasses a wide variety of formats, from full courses, textbooks, and curricula to maps, videos, podcasts, multimedia applications, assessment tools, mobile apps, databases, and even simulations.

Open educational resources are made up of three elements that work inseparably:

1. Educational content: includes all kinds of material that can be used in the teaching-learning process, from formal objects to external and social resources. This is where open data would come in, which can be used to generate this type of resource.
2. Technological tools: software that allows content to be developed, used, modified and distributed, including applications for content creation and platforms for learning communities.
3. Open licenses: differentiating element that respects intellectual property while providing permissions for the use, adaptation and redistribution of materials.

Therefore, OER are mainly characterized by their universal accessibility, eliminating economic and geographical barriers that traditionally limit access to quality education.

Educational innovation and pedagogical transformation

Pedagogical transformation is one of the main impacts of open educational resources in the current educational landscape. OER are not simply free digital content, but catalysts for innovation that are redefining teaching-learning processes globally.

Combined with appropriate pedagogical methodologies and well-designed learning objectives, OER offer innovative new teaching options to enable both teachers and students to take a more active role in the educational process and even in the creation of content. They foster essential competencies such as critical thinking, autonomy and the ability to "learn to learn", overcoming traditional models based on memorization.

Educational innovation driven by OER is materialized through open technological tools that facilitate their creation, adaptation and distribution. Programs such as eXeLearning allow you to develop digital educational content in a simple way, while LibreOffice and Inkscape offer free alternatives for the production of materials.

The interoperability achieved through open standards, such as IMS Global or SCORM, ensures that these resources can be integrated into different platforms and therefore accessibility for all users, including people with disabilities.

Another promising innovation for the future of OER is the combination of decentralized technologies like Nostr with authoring tools like LiaScript. This approach solves the dependency on central servers, allowing an entire course to be created and distributed over an open, censorship-resistant network. The result is a single, permanent link (URI de Nostr) that encapsulates all the material, giving the creator full sovereignty over its content and ensuring its durability. In practice, this is a revolution for universal access to knowledge. Educators share their work with the assurance that the link will always be valid, while students access the material directly, without the need for platforms or intermediaries. This technological synergy is a fundamental step to materialize the promise of a truly open, resilient and global educational ecosystem, where knowledge flows without barriers.

The potential of Open Educational Resources is realized thanks to the communities and projects that develop and disseminate them. Institutional initiatives, collaborative repositories and programmes promoted by public bodies and teachers ensure that OER are accessible, reusable and sustainable.

Collaboration and open learning communities

The collaborative dimension represents one of the fundamental pillars that support the open educational resources movement. This approach transcends borders and connects education professionals globally.

The educational communities around OER have created spaces where teachers share experiences, agree on methodological aspects and resolve doubts about the practical application of these resources. Coordination between professionals usually occurs on social networks or through digital channels such as Telegram, in which both users and content creators participate. This "virtual cloister" facilitates the effective implementation of active methodologies in the classroom.

Beyond the spaces that have arisen at the initiative of the teachers themselves, different organizations and institutions have promoted collaborative projects and platforms that facilitate the creation, access and exchange of Open Educational Resources, thus expanding their reach and impact on the educational community.

OER projects and repositories in Spain

In the case of Spain, Open Educational Resources have a consolidated ecosystem of initiatives that reflect the collaboration between public administrations, educational centres, teaching communities and cultural entities. Platforms such as Procomún, content creation projects such as EDIA (Educational, Digital, Innovative and Open) or CREA (Creation of Open Educational Resources), and digital repositories such as Hispana show the diversity of approaches adopted to make educational and cultural resources available to citizens in open access. Here's a little more about them:

• The EDIA (Educational, Digital, Innovative and Open) Project, developed by the National Center for Curriculum Development in Non-Proprietary Systems (CEDEC), focuses on the creation of open educational resources designed to be integrated into environments that promote digital competences and that are adapted to active methodologies. The resources are created with eXeLearning, which facilitates editing, and include templates, guides, rubrics and all the necessary documents to bring the didactic proposal to the classroom.
• The Procomún network  was born as a result of the Digital Culture in School Plan launched in 2012 by the Ministry of Education, Culture and Sport. This repository currently has more than 74,000 resources and 300 learning itineraries, along with a multimedia bank of 100,000 digital assets under the Creative Commons  license and which, therefore, can be reused to create new materials. It also has a mobile application. Procomún also uses eXeLearning and the LOM-ES standard, which ensures a homogeneous description of the resources and facilitates their search and classification. In addition, it is a semantic web, which means that it can connect with existing communities through the Linked Open Data Cloud.

• The autonomous communities have also promoted the creation of open educational resources. An example is CREA, a programme of the Junta de Extremadura aimed at the collaborative production of open educational resources. Its platform allows teachers to create, adapt and share structured teaching materials, integrating curricular content with active methodologies. The resources are generated in interoperable formats and are accompanied by metadata that facilitates their search, reuse and integration into different platforms.

  There are similar initiatives, such as the REA-DUA project  in Andalusia, which brings together more than 250 educational resources for primary, secondary and baccalaureate, with attention to diversity. For its part, Galicia launched the 2022-23 academic year cREAgal whose portal currently has more than 100 primary and secondary education resources. This project has an impact on inclusion and promotes the personal autonomy of students. In addition, some ministries of education make open educational resources available, as is the case of the Canary Islands.

Hispana, the portal for access to Spanish cultural heritage

In addition to these initiatives aimed at the creation of educational resources, others have emerged that promote the collection of content that was not created for an educational purpose but that can be used in the classroom. This is the case of Hispana, a portal for aggregating digital collections from Spanish libraries, archives and museums.

To provide access to Spanish cultural and scientific heritage, Hispana collects and makes accessible the metadata of digital objects, allowing these objects to be viewed through links to the pages of the owner institutions. In addition to acting as a collector, Hispana also adds the content of institutions that wish to do so to Europeana, the European digital library, which allows increasing the visibility and reuse of resources.

Hispana is an OAI-PMH repository, which means that it uses the Open Archives Initiative – Protocol for Metadata Harvesting, an international standard for the collection and exchange of metadata between digital repositories. Thus, Hispana collects the metadata of the Spanish archives, museums and libraries that exhibit their collections with this protocol and sends them to Europeana.

International initiatives and global cooperation

At the global level, it is important to highlight the role of UNESCO through the Dynamic Coalition on OER, which seeks to coordinate efforts to increase the availability, quality and sustainability of these assets.

In Europe, ENCORE+ (European Network for Catalysing Open Resources in Education) seeks to strengthen the European OER ecosystem. Among its objectives is to create a network that connects universities, companies and public bodies to promote the adoption, reuse and quality of OER in Europe. ENCORE+ also promotes interoperability between platforms, metadata standardization and cooperation to ensure the quality of resources.

In Europe, other interesting initiatives have been developed, such as EPALE (Electronic Platform for Adult Learning in Europe), an initiative of the European Commission aimed at specialists in adult education. The platform contains studies, reports and training materials, many of them under open licenses, which contributes to the dissemination and use of OER.

In addition, there are numerous projects that generate and make available open educational resources around the world. In the United States, OER Commons functions as a global repository of educational materials of different levels and subjects. This project uses Open Author, an online editor that makes it easy for teachers without advanced technical knowledge to create and customize digital educational resources directly on the platform.

Another outstanding project is Plan Ceibal, a public program in Uruguay that represents a model of technological inclusion for equal opportunities. In addition to providing access to technology, it generates and distributes OER in interoperable formats, compatible with standards such as SCORM and structured metadata that facilitate its search, integration into learning platforms and reuse by teachers.

Along with initiatives such as these, there are others that, although they do not directly produce open educational resources, do encourage their creation and use through collaboration between teachers and students from different countries. This is the case for projects such as eTwinning and Global Classroom.

The strength of OER lies in their contribution to the democratization of knowledge, their collaborative nature, and their ability to promote innovative methodologies. By breaking down geographical, economic, and social barriers, open educational resources bring the right to education one step closer to becoming a universal reality.

Open data from public sources has evolved over the years, from being simple repositories of information to constituting dynamic ecosystems that can transform public governance. In this context, artificial intelligence (AI) emerges as a catalytic technology that benefits from the value of open data and exponentially enhances its usefulness. In this post we will see what the mutually beneficial symbiotic relationship between AI and open data looks like.

Traditionally, the debate on open data has focused on portals: the platforms on which governments publish information so that citizens, companies and organizations can access it. But the so-called "Third Wave of Open Data," a term by New York University's GovLab, emphasizes that it is no longer enough to publish datasets on demand or by default. The important thing is to think about the entire ecosystem: the life cycle of data, its exploitation, maintenance and, above all, the value it generates in society.

What role can open data play in AI?

In this context, AI appears as a catalyst capable of automating tasks, enriching open government data (DMOs), facilitating its understanding and stimulating collaboration between actors.

Recent research, developed by European universities, maps how this silent revolution is happening. The study proposes a classification of uses according to two dimensions:

1. Perspective, which in turn is divided into two possible paths:

   1. Inward-looking (portal): The focus is on the internal functions of data portals.

   2. Outward-looking (ecosystem): the focus is extended to interactions with external actors (citizens, companies, organizations).

2. Phases of the data life cycle, which can be divided into pre-processing, exploration, transformation and maintenance.

In summary, the report identifies these eight types of AI use in government open data, based on perspective and phase in the data lifecycle.

Figure 1. Eight uses of AI to improve government open data. Source: presentation “Data for AI or AI for data: artificial intelligence as a catalyst for open government ecosystems”, based on the report of the same name, from EU Open Data Days 2025.

A continuación, se detalla cada uno de estos usos:

1. Portal curator

This application focuses on pre-processing data within the portal. AI helps organize, clean, anonymize, and tag datasets before publication. Some examples of tasks are:

• Automation and improvement of data publication tasks.

• Performing auto-tagging and categorization functions.

• Data anonymization to protect privacy.

• Automatic cleaning and filtering of datasets.

• Feature extraction and missing data handling.

2. Ecosystem data retriever

Also in the pre-processing phase, but with an external focus, AI expands the coverage of portals by identifying and collecting information from diverse sources. Some tasks are:

• Retrieve structured data from legal or regulatory texts.

• News mining to enrich datasets with contextual information.

• Integration of urban data from sensors or digital records.

• Discovery and linking of heterogeneous sources.

• Conversion of complex documents into structured information.

3. Portal explorer

In the exploration phase, AI systems can also make it easier to find and interact with published data, with a more internal approach. Some use cases:

• Develop semantic search engines to locate datasets.

• Implement chatbots that guide users in data exploration.

• Provide natural language interfaces for direct queries.

• Optimize the portal's internal search engines.

• Use language models to improve information retrieval.

4. Ecosystem connector

Operating also in the exploration phase, AI acts as a bridge between actors and ecosystem resources. Some examples are:

• Recommend relevant datasets to researchers or companies.

• Identify potential partners based on common interests.

• Extract emerging themes to support policymaking.

• Visualize data from multiple sources in interactive dashboards.

• Personalize data suggestions based on social media activity.

5. Portal linker

This functionality focuses on the transformation of data within the portal. Its function is to facilitate the combination and presentation of information for different audiences. Some tasks are:

• Convert data into knowledge graphs (structures that connect related information, known as Linked Open Data).

• Summarize and simplify data with NLP (Natural Language Processing) techniques.

• Apply automatic reasoning to generate derived information.

• Enhance multivariate visualization of complex datasets.

• Integrate diverse data into accessible information products.

6. Ecosystem value developer

In the transformation phase and with an external perspective, AI generates products and services based on open data that provide added value. Some tasks are:

• Suggest appropriate analytical techniques based on the type of dataset.

• Assist in the coding and processing of information.

• Create dashboards based on predictive analytics.

• Ensure the correctness and consistency of the transformed data.

• Support the development of innovative digital services.

7. Portal monitor

It focuses on portal maintenance, with an internal focus. Their role is to ensure quality, consistency, and compliance with standards. Some tasks are:

• Detect anomalies and outliers in published datasets.

• Evaluate the consistency of metadata and schemas.

• Automate data updating and purification processes.

• Identify incidents in real time for correction.

• Reduce maintenance costs through intelligent monitoring.

8. Ecosystem engager

And finally, this function operates in the maintenance phase, but outwardly. It seeks to promote citizen participation and continuous interaction. Some tasks are:

• Predict usage patterns and anticipate user needs.

• Provide personalized feedback on datasets.

• Facilitate citizen auditing of data quality.

• Encourage participation in open data communities.

• Identify user profiles to design more inclusive experiences.

What does the evidence tell us?

The study is based on a review of more than 70 academic papers examining the intersection between AI and OGD (open government data). From these cases, the authors observe that:

• Some of the defined profiles, such as portal curator, portal explorer and portal monitor, are relatively mature and have multiple examples in the literature.

• Others, such as ecosystem value developer and ecosystem engager, are less explored, although they have the most potential to generate social and economic impact.

• Most applications today focus on automating specific tasks, but there is a lot of scope to design more comprehensive architectures, combining several types of AI in the same portal or across the entire data lifecycle.

From an academic point of view, this typology provides a common language and conceptual structure to study the relationship between AI and open data. It allows identifying gaps in research and guiding future work towards a more systemic approach.

In practice, the framework is useful for:

• Data portal managers: helps them identify what types of AI they can implement according to their needs, from improving the quality of datasets to facilitating interaction with users.

• Policymakers: guides them on how to design AI adoption strategies in open data initiatives, balancing efficiency, transparency, and participation.

• Researchers and developers: it offers them a map of opportunities to create innovative tools that address specific ecosystem needs.

Limitations and next steps of the synergy between AI and open data

In addition to the advantages, the study recognizes some pending issues that, in a way, serve as a roadmap for the future. To begin with, several of the applications that have been identified are still in early stages or are conceptual. And, perhaps most relevantly, the debate on the risks and ethical dilemmas of the use of AI in open data has not yet been addressed in depth: bias, privacy, technological sustainability.

In short, the combination of AI and open data is still a field under construction, but with enormous potential. The key will be to move from isolated experiments to comprehensive strategies, capable of generating social, economic and democratic value. AI, in this sense, does not work independently of open data: it multiplies it and makes it more relevant for governments, citizens and society in general.

We know that the open data managed by the public sector in the exercise of its functions is an invaluable resource for promoting transparency, driving innovation and stimulating economic development. At the global level, in the last 15 years this idea has led to the creation of data portals that serve as  a single point of access for public information both in a country, a region or city.

However, we sometimes find that the full exploitation of the potential of open data is limited by problems inherent in its quality. Inconsistencies, lack of standardization or interoperability, and incomplete metadata are just some of the common challenges that sometimes undermine the usefulness of open datasets and that government agencies also point to as the main obstacle to AI adoption.

When we talk about the relationship between open data and artificial intelligence, we almost always start from the same idea: open data feeds AI, that is, it is part of the fuel for models. Whether it's to train foundational models like ALIA, to specialize small language models (SLMs) versus LLMs, or to evaluate and validate their capabilities or explain their behavior (XAI), the argument revolves around the usefulness of open data for artificial intelligence, forgetting that open data was already there and has many other uses.

Therefore, we are going to reverse the perspective and explore how AI itself can become a powerful tool to improve the quality and, therefore, the value of open data itself. This approach, which was already outlined  by the United Nations Economic Commission for Europe (UNECE) in its pioneering  2022 Machine Learning for Official Statistics report  , has become more relevant since the explosion of generative AI. We can now use the artificial intelligence available to increase the quality of datasets that are published throughout their entire lifecycle: from capture and normalization to validation, anonymization, documentation, and follow-up in production.

With this, we can increase the public value of data, contribute to its reuse and amplify its social and economic impact. And, at the same time, to improve the quality of the next generation of artificial intelligence models.

Common challenges in open data quality

Data quality has traditionally been a Critical factor for the success of any open data initiative, which is cited in numerous reports such as that of the European Commission "Improving data publishing by open data portal managers and owners”. The most frequent challenges faced by data publishers include:

• Inconsistencies and errors: Duplicate data, heterogeneous formats, or outliers are common in datasets. Correcting these small errors, ideally at the data source itself, was traditionally costly and greatly limited the usefulness of many datasets.

• Lack of standardization and interoperability: Two sets that talk about the same thing may name columns differently, use non-comparable classifications, or lack persistent identifiers to link entities. Without a common minimum, combining sources becomes an artisanal work that makes it more expensive to reuse data.

• Incomplete or inaccurate metadata: The lack of clear information about the origin, collection methodology, frequency of updating or meaning of the fields, complicates the understanding and use of the data. For example, knowing with certainty if the resource can be integrated into a service, if it is up to date or if there is a point of contact to resolve doubts is very important for its reuse.
• Outdated or outdated data: In highly dynamic domains such as mobility, pricing, or environmental data, an outdated set can lead to erroneous conclusions. And if there are no versions, changelogs, or freshness indicators, it's hard to know what's changed and why. The absence of a "history" of the data complicates auditing and reduces trust.
• Inherent biases: sometimes coverage is incomplete, certain populations are underrepresented, or a management practice introduces systematic deviation. If these limits are not documented and warned, analyses can reinforce inequalities or reach unfair conclusions without anyone noticing.

Where Artificial Intelligence Can Help

Fortunately, in its current state, artificial intelligence is already in a position to provide a set of tools that can help address some of these open data quality challenges, transforming your management from a manual and error-prone process to a more automated and efficient one:

• Automated error detection and correction: Machine learning algorithms and AI models can automatically and reliably identify inconsistencies, duplicates, outliers, and typos in large volumes of data. In addition, AI can help normalize and standardize data, transforming it for example into common formats and schemas to facilitate interoperability (such as DCAT-AP), and at a fraction of the cost it was so far.
• Metadata enrichment and cataloging: Technologies associated with natural language processing (NLP), including the use of large language models (LLMs) and small language models (SLMs), can help analyze descriptions and generate more complete and accurate metadata. This includes tasks such as suggesting relevant tags, classification categories, or extracting key entities (place names, organizations, etc.) from textual descriptions to enrich metadata.
• Anonymization and privacy: When open data contains information that could affect privacy, anonymization becomes a critical, but sometimes costly, task. Artificial Intelligence can contribute to making anonymization much more robust and to minimize risks related to re-identification by combining different data sets.

Bias assessment: AI can analyze the open datasets themselves for representation or historical biases. This allows publishers to take steps to correct them or at least warn users about their presence so that they are taken into account when they are to be reused. In short, artificial intelligence should not only be seen as a "consumer" of open data, but also as a strategic ally to improve its quality. When integrated with standards, processes, and human oversight, AI helps detect and explain incidents, better document sets, and publish trust-building quality evidence. As described in the 2024 Artificial Intelligence Strategy, this synergy unlocks more public value: it facilitates innovation, enables better-informed decisions, and consolidates a more robust and reliable open data ecosystem with more useful, more reliable open data with greater social impact.

In addition, a virtuous cycle is activated: higher quality open data trains more useful and secure models; and more capable models make it easier to continue raising the quality of data. In this way, data management is no longer a static task of publication and becomes a dynamic process of continuous improvement.

Content created by Jose Luis Marín, Senior Consultant in Data, Strategy, Innovation & Digitalisation. The content and views expressed in this publication are the sole responsibility of the author.

Artificial intelligence (AI) has become a central technology in people's lives and in the strategy of companies. In just over a decade, we've gone from interacting with virtual assistants that understood simple commands, to seeing systems capable of writing entire reports, creating hyper-realistic images, or even writing code.

This visible leap has made many wonder: is it all the same? What is the difference between what we already knew as AI and this new "Generative AI" that is so much talked about?

In this article we are going to organize those ideas and explain, with clear examples, how "Traditional" AI and Generative AI fit  under the great umbrella of artificial intelligence.

Traditional AI: analysis and prediction

For many years, what we understood by AI was closer to what we now call "Traditional AI". These systems are characterized by solving concrete, well-defined problems within a framework of available rules or data.

Some practical examples:

• Recommendation engines: Spotify suggests songs based on your listening history and Netflix adjusts its catalog to your personal tastes, generating up to 80% of views on the platform.

• Prediction systems: Walmart uses predictive models to anticipate the demand for products based on factors such as weather or local events; Red Eléctrica de España applies similar algorithms to forecast electricity consumption and balance the grid.

• Automatic recognition: Google Photos classifies images by recognizing faces and objects; Visa and Mastercard use anomaly detection models to identify fraud in real time; Tools like Otter.ai automatically transcribe meetings and calls.

In all these cases, the models learn from past data to provide a classification, prediction, or decision. They do not invent anything new, but recognize patterns and apply them to the future.

Generative AI: content creation

The novelty of generative AI is that it not only analyzes, but also produces (generates) from the data it has.

In practice, this means that:

• You can generate structured text from a couple of initial ideas.

• You can combine existing visual elements from a written description.

• You can create product prototypes, draft presentations, or propose code snippets based on learned patterns.

The key is that generative models don't just classify or predict, they generate new combinations based on what they learned during their training.

The impact of this breakthrough is enormous: in the development world, GitHub Copilot already includes agents that detect and fix programming errors on their own; in design, Google's Nano Banana  tool  promises to revolutionize image editing with an efficiency that could render programs like Photoshop obsolete; and in music, entirely AI-created bands like Velvet Velvet Sundown they already exceed one million monthly listeners on Spotify, with songs, images and biography fully generated, without real musicians behind them.

When is it best to use each type of AI?

The choice between Traditional and Generative AI is not a matter of fashion, but of what specific need you want to solve. Each shines in different situations:

Traditional AI: the best option when...

• You need to predict future behaviors based on historical data (sales, energy consumption, predictive maintenance).

• You want to detect anomalies or classify information accurately (transaction fraud, imaging, spam).

• You are looking  to optimize processes to gain efficiency (logistics, transport routes, inventory management).

• You work in critical environments where reliability and accuracy are a must (health, energy, finance).

Use it when the goal is to make decisions based on real data with the highest possible accuracy.

Generative AI: the best option when...

• You need to create content (texts, images, music, videos, code).

• You want to prototype or experiment quickly, exploring different scenarios before deciding (product design, R+D testing).

• You are looking for more natural interaction with users (chatbots, virtual assistants, conversational interfaces).

• You require large-scale personalization, generating messages or materials adapted to each individual (marketing, training, education).

• You are interested in simulating scenarios that you cannot easily obtain with real data (fictitious clinical cases, synthetic data to train other models).

Use it when the goal is to create, personalize, or interact in a more human and flexible way.

An example from the health field illustrates this well:

• Traditional AI can analyze thousands of clinical records to anticipate the likelihood of a patient developing a disease.

• Generative AI can create fictional scenarios to train medical students, generating realistic clinical cases without exposing real patient data.

Do they compete or complement each other?

In 2019, Gartner introduced the concept of Composite AI to describe hybrid solutions that combined different AI approaches to solve a problem more comprehensively. Although it was a term that was not very widespread then, today it is more relevant than ever thanks to the emergence of Generative AI.

Generative AI does not replace Traditional AI, but rather complements it. When you integrate both approaches into a single workflow, you achieve much more powerful results than if you used each technology separately.

Although, according to Gartner, Composite AI is still in the Innovation Trigger phase, where an emerging technology begins to generate interest, and although its practical use is still limited, we already see many new trends being generated in multiple sectors:

• In retail: A traditional system predicts how many orders a store will receive next week, and generative AI automatically generates personalized product descriptions for customers of those orders.

• In education: a traditional model assesses student progress and detects weak areas, while generative AI designs exercises or materials tailored to those needs.

• In industrial design: a traditional algorithm optimizes manufacturing logistics, while a generative AI proposes prototypes of new parts or products.

Ultimately, instead of questioning which type of AI is more advanced, the right thing to do is to ask: what problem do I want to solve, and which AI approach is right for it?

Content created by Juan Benavente, senior industrial engineer and expert in technologies related to the data economy. The content and views expressed in this publication are the sole responsability of the author. 

From today, September 15, registration is open for one of the most important events in the geospatial sector in the Iberian Peninsula. The XVI Iberian Conference on Spatial Data Infrastructures (JIIDE 2025) will be held in Oviedo from 12 to 14 November 2025. This annual meeting represents a unique opportunity to explore the latest trends in spatial data reuse, especially in the context of the application of artificial intelligence to territorial knowledge.

Since its first edition in 2011, the JIIDEs have evolved as a result of collaboration between the Direção-Geral do Território de Portugal, the National Geographic Institute of Spain through the National Center for Geographic Information, and the Government of Andorra. In this sixteenth edition, the Ministry of Territorial Planning, Urban Planning, Housing and Citizens' Rights of  the Principality of Asturias and the University of Oviedo also join, thus consolidating an initiative that brings together hundreds of professionals from the Public Administration, the private sector and the academic field every year.

For three days, experts with proven experience and technical knowledge in geographic information will share their most innovative developments, work methodologies and success stories in the management and reuse of spatial data.

Two axes: artificial intelligence and the INSPIRE and HVDS regulatory framework

The central theme of this edition, "AI and territory: exploring the new frontiers of spatial knowledge", reflects the natural evolution of the sector towards the incorporation of emerging technologies. Artificial intelligence, machine learning, and advanced analytics algorithms are radically transforming the way we process, analyze, and extract value from geospatial data.

This orientation towards AI is not accidental. The publication and use of geospatial data makes it possible to harness one of the most valuable digital assets for economic development, environmental monitoring, competitiveness, innovation and job creation. When this data is combined with artificial intelligence techniques, its potential multiplies exponentially.

The conference takes place at a particularly relevant time for the open data ecosystem. The INSPIRE Directive, together with Directive (EU) 2019/1024 on open data and re-use of public sector information, has established a regulatory framework that explicitly recognises the economic and social value of digital geospatial data.

The evolution in the publication of high-value datasets marks an important milestone in this process. These sets, characterized by their great potential for reuse, should be available free of charge, in machine-readable formats and through application programming interfaces (APIs). Geospatial data occupies a central position in this categorisation, underlining its strategic importance for the European open data ecosystem.

JIIDE 2025 will devote particular attention to presenting practical examples of re-use of these high-value datasets , both through the new OGC APIs and through traditional download services and established interoperable formats. This practical approach will allow attendees to learn about real cases of implementation and their tangible results.

Miscellaneous Program: Use Cases, AI, and Geospatial Data Reuse

You can also check the program here. Among the planned activities, there are sessions ranging from fundamental technical aspects to innovative applications that demonstrate the transformative potential of this data. The activities are organized into five main themes:

1. Spatial data structure and metadata.

2. Data management and publication.

3. Development of spatial  software.

4. Artificial intelligence.

5. Cooperation between agents.

Some of the highlighted topics are project management and coordination, where corporate systems such as the SIG of the Junta de Andalucía or the SITNA of the Government of Navarra will be presented. Earth observation will also feature prominently, with presentations on the evolution of the National Plan for Aerial Orthophotography (APNOA) programme  and advanced deep learning image processing techniques.

On the other hand, thematic visualisers also represent another fundamental axis, showing how spatial data can be transformed into accessible tools for citizens. From eclipse visualizers to tools for calculating the solar potential of rooftops, developments will be presented that demonstrate how the creative reuse of data can generate services of high social value.

Following the annual theme, the application of AI to geospatial data will be approached from multiple perspectives. Use cases will be presented in areas as diverse as the automatic detection of sports facilities, the classification of LiDAR point clouds, the identification of hazardous materials such as asbestos, or the optimization of urban mobility.

One of the most relevant sessions for the open data community will focus specifically on "Reuse and Open Government". This session will address the integration of spatial data infrastructures into open data portals, spatial data metadata according to the GeoDCAT-AP standard, and the application of data quality regulations.

Local governments play a key role in the generation and publication of spatial data. For this reason, the JIIDE 2025 will dedicate a specific session to the publication of local data, where municipalities such as Barcelona, Madrid, Bilbao or Cáceres will share their experiences and developments.

In addition to the theoretical sessions, the conferences include practical workshops on specific tools, methodologies and technologies. These workshops, lasting 45 minutes to an hour, allow attendees to experiment directly with the solutions presented. Some of them address the creation of custom web geoportals and others, for example, the implementation of OGC APIs, through advanced visualization techniques and metadata management tools.

Participate in person or online

The JIIDEs maintain their commitment to open participation, inviting both researchers and professionals to present their tools, technical solutions, work methodologies and success stories. In addition, the JIIDE 2025 will be held in hybrid mode, allowing both face-to-face participation in Oviedo and virtual monitoring.

This flexibility, maintained from the experiences of recent years, ensures that professionals throughout the Iberian territory and beyond can benefit from shared knowledge. Participation remains free, although prior registration is required for each session, roundtable or workshop.

Starting today, you can sign up and take advantage of this opportunity to learn and exchange experiences on geospatial data. Registration is available on the official website of the event: https://www.jiide.org/web/portal/inicio

Synthetic images are visual representations artificially generated by algorithms and computational techniques, rather than being captured directly from reality with cameras or sensors. They are produced from different methods, among which the antagonistic generative networks (Generative Adversarial NetworksGAN), the Dissemination models, and the 3D rendering techniques. All of them allow you to create images of realistic appearance that in many cases are indistinguishable from an authentic photograph.

When this concept is transferred to the field of Earth observation, we are talking about synthetic satellite images. These are not obtained from a space sensor that captures real electromagnetic radiation, but are generated digitally to simulate what a satellite would see from orbit. In other words, instead of directly reflecting the physical state of the terrain or atmosphere at a particular time, they are computational constructs capable of mimicking the appearance of a real satellite image.

The development of this type of image responds to practical needs. Artificial intelligence systems that process remote sensing data require very large and varied sets of images. Synthetic images allow, for example, to recreate areas of the Earth that are little observed, to simulate natural disasters – such as forest fires, floods or droughts – or to generate specific conditions that are difficult or expensive to capture in practice. In this way, they constitute a valuable resource for training detection and prediction algorithms in agriculture, emergency management, urban planning or environmental monitoring.

 

 

Its value is not limited to model training. Where high-resolution images do not exist – due to technical limitations, access restrictions or economic reasons – synthesis makes it possible to fill information gaps and facilitate preliminary studies. For example, researchers can work with approximate synthetic images to design risk models or simulations before actual data are available.

However, synthetic satellite imagery also poses significant risks. The possibility of generating very realistic scenes opens the door to manipulation and misinformation. In a geopolitical context, an image showing non-existent troops or destroyed infrastructure could influence strategic decisions or international public opinion. In the environmental field, manipulated images could be disseminated to exaggerate or minimize the impacts of phenomena such as deforestation or melting ice, with direct effects on policies and markets.

Therefore, it is convenient to differentiate between two very different uses. The first is use as a support, when synthetic images complement real images to train models or perform simulations. The second is use as a fake, when they are deliberately presented as authentic images in order to deceive. While the former uses drive innovation, the latter threatens trust in satellite data and poses an urgent challenge of authenticity and governance.

Risks of satellite imagery applied to Earth observation

Synthetic satellite imagery poses significant risks when used in place of images captured by real sensors. Below are examples that demonstrate this.

A new front of disinformation: "deepfake geography"

The term deepfake geography has already been consolidated in the academic and popular literature to describe fictitious satellite images, manipulated with AI, that appear authentic, but do not reflect any existing reality. Research from the University of Washington, led by Bo Zhao, used algorithms such as CycleGAN to modify images of real cities—for example, altering the appearance of Seattle with non-existent buildings or transforming Beijing into green areas—highlighting the potential to generate convincing false landscapes.

One OnGeo Intelligence (OGC) platform article stresses that these images are not purely theoretical, but real threats affecting national security, journalism and humanitarian work. For its part, the OGC warns that fabricated satellite imagery, AI-generated urban models, and synthetic road networks have already been observed, and that they pose real challenges to public and operational trust.

Strategic and policy implications

Satellite images are considered "impartial eyes" on the planet, used by governments, media and organizations. When these images are faked, their consequences can be severe:

• National security and defense: if false infrastructures are presented or real ones are hidden, strategic analyses can be diverted or mistaken military decisions can be induced.
• Disinformation in conflicts or humanitarian crises: An altered image showing fake fires, floods, or troop movements can alter the international response, aid flows, or citizens' perceptions, especially if it is spread through social media or media without verification.
• Manipulation of realistic images of places: not only the general images are at stake. Nguyen et al. (2024) showed that it is possible to generate highly realistic synthetic satellite images of very specific facilities such as nuclear plants.

Crisis of trust and erosion of truth

For decades, satellite imagery has been perceived as one of the most objective and reliable sources of information about our planet. They were the graphic evidence that made it possible to confirm environmental phenomena, follow armed conflicts or evaluate the impact of natural disasters. In many cases, these images were used as "unbiased evidence," difficult to manipulate, and easy to validate. However, the emergence of synthetic images generated by artificial intelligence has begun to call into question that almost unshakable trust.

Today, when a satellite image can be falsified with great realism, a profound risk arises: the erosion of truth and the emergence of a crisis of confidence in spatial data.

The breakdown of public trust

When citizens can no longer distinguish between a real image and a fabricated one, trust in information sources is broken. The consequence is twofold:

• Distrust of institutions: if false images of a fire, a catastrophe or a military deployment circulate and then turn out to be synthetic, citizens may also begin to doubt the authentic images published by space agencies or the media. This "wolf is coming" effect generates skepticism even in the face of legitimate evidence.
• Effect on journalism: traditional media, which have historically used satellite imagery as an unquestionable visual source, risk losing credibility if they publish doctored images without verification. At the same time, the abundance of fake images on social media erodes the ability to distinguish what is real and what is not.
• Deliberate confusion: in contexts of disinformation, the mere suspicion that an image may be false can already be enough to generate doubt and sow confusion, even if the original image is completely authentic.

The following is a summary of the possible cases of manipulation and risk in satellite images:

Ambit	Type of handling	Main risk	Documented example
Armed conflicts	Insertion or elimination of military infrastructures.	Strategic disinformation; erroneous military decisions; loss of credibility in international observation.	Alterations demonstrated in deepfake geography studies  where dummy roads, bridges or buildings were added to satellite images.
Climate change and the environment	Alteration of glaciers, deforestation or emissions.	Manipulation of environmental policies; delay in measures against climate change; denialism.	Studies have shown the ability to generate modified landscapes (forests in urban areas, changes in ice) by means of GANs.
Gestión de emergencias	Creation of non-existent disasters (fires, floods).	Misuse of resources in emergencies; chaos in evacuations; loss of trust in agencies.	Research has shown the ease of inserting smoke, fire or water into satellite images.
Mercados y seguros	Falsification of damage to infrastructure or crops.	Financial impact; massive fraud; complex legal litigation.	Potential use of fake images to exaggerate damage after disasters and claim compensation or insurance.
Derechos humanos y justicia internacional	Alteration of visual evidence of war crimes.	Delegitimization of international tribunals; manipulation of public opinion.	Risk identified in intelligence reports: Doctored images could be used to accuse or exonerate actors in conflicts.
Geopolítica y diplomacia	Creation of fictitious cities or border changes.	Diplomatic tensions; treaty questioning; State propaganda	Examples of deepfake maps that transform geographical features of cities such as Seattle or Tacoma.

Figure 2. Table showing possible cases of manipulation and risk in satellite images

Impact on decision-making and public policies

The consequences of relying on doctored images go far beyond the media arena:

• Urbanism and planning: decisions about where to build infrastructure or how to plan urban areas could be made on manipulated images, generating costly errors that are difficult to reverse.
• Emergency management: If a flood or fire is depicted in fake images, emergency teams can allocate resources to the wrong places, while neglecting areas that are actually affected.
• Climate change and the environment: Doctored images of glaciers, deforestation or polluting emissions could manipulate political debates and delay the implementation of urgent measures.
• Markets and insurance: Insurers and financial companies that rely on satellite imagery to assess damage could be misled, with significant economic consequences.

In all these cases, what is at stake is not only the quality of the information, but also the effectiveness and legitimacy of public policies based on that data.

The technological cat and mouse game

The dynamics of counterfeit generation and detection are already known in other areas, such as  video or audio deepfakes: every time a more realistic generation method emerges, a more advanced detection algorithm is developed, and vice versa. In the field of satellite images, this technological career has particularities:

• Increasingly sophisticated generators: today's broadcast models can create highly realistic scenes, integrating ground textures, shadows, and urban geometries that fool even human experts.
• Detection limitations: Although algorithms are developed to identify fakes (analyzing pixel patterns, inconsistencies in shadows, or metadata), these methods are not always reliable when faced with state-of-the-art generators.
• Cost of verification: independently  verifying a satellite image requires access to alternative sources or different sensors, something that is not always available to journalists, NGOs or citizens.
• Double-edged swords: The same techniques used to detect fakes can be exploited by those who generate them, further refining synthetic images and making them more difficult to differentiate.

From visual evidence to questioned evidence

The deeper impact is cultural and epistemological: what was previously assumed to be objective evidence now becomes an element subject to doubt. If satellite imagery is no longer perceived as reliable evidence, it weakens fundamental narratives around scientific truth, international justice, and political accountability.

• In armed conflicts, a satellite image showing possible war crimes can be dismissed under the accusation of being a deepfake.
• In international courts, evidence based on satellite observation could lose weight in the face of suspicion of manipulation.
• In public debate, the relativism of "everything can be false" can be used as a rhetorical weapon to delegitimize even the strongest evidence.

Strategies to ensure authenticity

The crisis of confidence in satellite imagery is not an isolated problem in the geospatial sector, but is part of a broader phenomenon: digital disinformation in the age of artificial intelligence. Just as  video deepfakes have called into question the validity of audiovisual evidence, the proliferation of synthetic satellite imagery threatens to weaken the last frontier of perceived objective data: the unbiased view from space.

Ensuring the authenticity of these images requires a combination of technical solutions and governance mechanisms, capable of strengthening traceability, transparency and accountability across the spatial data value chain. The main strategies under development are described below.

Robust metadata: Record origin and chain of custody

 Metadata is the first line of defense against manipulation. In satellite imagery, they should include detailed information about:

• The sensor used (type, resolution, orbit).
• The exact time of acquisition (date and time, with time precision).
• The precise geographical location (official reference systems).
• The applied processing chain (atmospheric corrections, calibrations, reprojections).

Recording this metadata in secure repositories allows the chain of custody to be reconstructed, i.e. the history of who, how and when an image has been manipulated. Without this traceability, it is impossible to distinguish between authentic and counterfeit images.

EXAMPLE: The  European Union's Copernicus program  already implements standardized and open metadata for all its Sentinel images, facilitating subsequent audits and confidence in the origin.

Digital signatures and blockchain: ensuring integrity

Digital signatures allow you to verify that an image has not been altered since it was captured. They function as a cryptographic seal that is applied at the time of acquisition and validated at each subsequent use.

Blockchain technology  offers an additional level of assurance: storing acquisition and modification records on an immutable chain of blocks. In this way, any changes in the image or its metadata would be recorded and easily detectable.

EXAMPLE: The ESA – Trusted Data Framework project  explores the use of blockchain to protect the integrity of Earth observation data and bolster trust in critical applications such as climate change and food security.

 Invisible watermarks: hidden signs in the image

Digital watermarking involves embedding imperceptible signals in the satellite image itself, so that any subsequent alterations can be detected automatically.

• It can be done at the pixel level, slightly modifying color patterns or luminance.
• It is combined with cryptographic techniques to reinforce its validity.
• It allows you to validate images even if they have been cropped, compressed, or reprocessed.

EXAMPLE: In the audiovisual sector, watermarks have been used for years in the protection of digital content. Its adaptation to satellite images is in the experimental phase, but it could become a standard verification tool.

Open Standards (OGC, ISO): Trust through Interoperability

Standardization is key to ensuring that technical solutions are applied in a coordinated and global manner.

• OGC (Open Geospatial Consortium) works on standards for metadata management, geospatial data traceability, and interoperability between systems. Their work on geospatial APIs and FAIR (Findable, Accessible, Interoperable, Reusable) metadata is essential to establishing common trust practices.
• ISO develops standards on information management and authenticity of digital records that can also be applied to satellite imagery.

EXAMPLE: OGC Testbed-19 included specific experiments on geospatial data authenticity, testing approaches such as digital signatures and certificates of provenance.

Cross-check: combining multiple sources

A basic principle for detecting counterfeits is to contrast sources. In the case of satellite imagery, this involves:

• Compare images from different satellites (e.g. Sentinel-2 vs. Landsat-9).
• Use different types of sensors (optical, radar SAR, hyperspectral).
• Analyze time series to verify consistency over time.

EXAMPLE: Damage verification in Ukraine following the start of the Russian invasion in 2022 was done by comparing images from several vendors (Maxar, Planet, Sentinel), ensuring that the findings were not based on a single source.

AI vs. AI: Automatic Counterfeit Detection

The same artificial intelligence that allows synthetic images to be created can be used to detect them. Techniques include:

• Pixel Forensics: Identify patterns generated by GANs or broadcast models.
• Neural networks trained to distinguish between real and synthetic images based on textures or spectral distributions.
• Geometric inconsistencies models: detect impossible shadows, topographic inconsistencies, or repetitive patterns.

EXAMPLE: Researchers at the University of Washington and other groups have shown that specific algorithms can detect satellite fakes with greater than 90% accuracy under controlled conditions.

Current Experiences: Global Initiatives

Several international projects are already working on mechanisms to reinforce authenticity:

• Coalition for Content Provenance and Authenticity (C2PA): A partnership between Adobe, Microsoft, BBC, Intel, and other organizations to develop an open standard for provenance and authenticity of digital content, including images. Its model can be applied directly to the satellite sector.
• OGC work: the organization promotes the debate on trust in geospatial data and has highlighted the importance of ensuring the traceability of synthetic and real satellite images (OGC Blog).
• NGA (National Geospatial-Intelligence Agency) in the US has publicly acknowledged the threat of synthetic imagery in defence and is driving collaborations with academia and industry to develop detection systems.

Towards an ecosystem of trust

The strategies described should not be understood as alternatives, but as complementary layers in a trusted ecosystem:

Id	Layers	Benefits
1	Robust metadata  (source, sensor, chain of custody)	Traceability guaranteed
2	Digital signatures and blockchain (data integrity)	Ensuring integrity
3	Invisible watermarks  (hidden signs)	Add a hidden level of protection
4	Cross-check  (multiple satellites and sensors)	Validates independently
5	AI vs. AI (counterfeit detector)	Respond to emerging threats
6	International governance  (accountability, legal frameworks)	Articulate clear rules of liability

Figure 3. Layers to ensure confidence in synthetic satellite images

Success will depend on these mechanisms being integrated together, under open and collaborative frameworks, and with the active involvement of space agencies, governments, the private sector and the scientific community.

Conclusions

Synthetic images, far from being just a threat, represent a powerful tool that, when used well, can provide significant value in areas such as simulation, algorithm training or innovation in digital services. The problem arises when these images are presented as real without proper transparency, fueling misinformation or manipulating public perception.

The challenge, therefore, is twofold: to take advantage of the opportunities offered by the synthesis of visual data to advance science, technology and management, and to minimize the risks associated with the misuse of these capabilities, especially in the form of deepfakes or deliberate falsifications.

In the particular case of satellite imagery, trust takes on a strategic dimension. Critical decisions in national security, disaster response, environmental policy, and international justice depend on them. If the authenticity of these images is called into question, not only the reliability of the data is compromised, but also the legitimacy of decisions based on them.

The future of Earth observation will be shaped by our ability to ensure authenticity, transparency and traceability across the value chain: from data acquisition to dissemination and end use. Technical solutions (robust metadata, digital signatures, blockchain, watermarks, cross-verification, and AI for counterfeit detection), combined with governance frameworks and international cooperation, will be the key to building an ecosystem of trust.

In short, we must assume a simple but forceful guiding principle:

"If we can't trust what we see from space, we put our decisions on Earth at risk."

Content prepared by Mayte Toscano, Senior Consultant in Data Economy Technologies. The contents and points of view reflected in this publication are the sole responsibility of the author.

Artificial intelligence (AI) assistants are already part of our daily lives: we ask them the time, how to get to a certain place or we ask them to play our favorite song. And although AI, in the future, may offer us infinite functionalities, we must not forget that linguistic diversity is still a pending issue.

In Spain, where Spanish coexists with co-official languages such as Basque, Catalan, Valencian and Galician, this issue is especially relevant. The survival and vitality of these languages in the digital age depends, to a large extent, on their ability to adapt and be present in emerging technologies. Currently, most virtual assistants, automatic translators or voice recognition systems do not understand all the co-official languages. However, did you know that there are collaborative projects to ensure linguistic diversity?

In this post we tell you about the approach and the greatest advances of some initiatives that are building the digital foundations necessary for the co-official languages in Spain to also thrive in the era of artificial intelligence.

ILENIA, the coordinator of multilingual resource initiatives in Spain

The models that we are going to see in this post share a focus because they are part of ILENIA, a state-level coordinator that connects the individual efforts of the autonomous communities. This initiative brings together the projects BSC-CNS (AINA), CENID (VIVES), HiTZ (NEL-GAITU) and the University of Santiago de Compostela (NÓS), with the aim of generating digital resources that allow the development of multilingual applications in the different languages of Spain.

The success of these initiatives depends fundamentally on citizen participation. Through platforms such as  Mozilla's Common Voice, any speaker can contribute to the construction of these linguistic resources through different forms of collaboration:

• Spoken Read: Collecting different ways of speaking through voice donations of a specific text.
• Spontaneous speech: creates  real and organic datasets as a result of conversations with prompts.
• Text in language: collaborate in the transcription of audios or in the contribution of textual content, suggesting new phrases or questions to enrich the corpora.

All resources are published under free licenses such as CC0, allowing them to be used free of charge by researchers, developers and companies.

The challenge of linguistic diversity in the digital age

Artificial intelligence systems learn from the data they receive during their training. To develop technologies that work correctly in a specific language, it is essential to have large volumes of data: audio recordings, text corpora and examples of real use of the language.

In other publications of datos.gob.es we have addressed the functioning of foundational models and initiatives in Spanish such as ALIA, trained with large corpus of text such as those of the Royal Spanish Academy.

Both posts explain why language data collection is not a cheap or easy task. Technology companies have invested massively in compiling these resources for languages with large numbers of speakers, but Spanish co-official languages face a structural disadvantage. This has led to many models not working properly or not being available in Valencian, Catalan, Basque or Galician.

However, there are collaborative and open data initiatives that allow the creation of quality language resources. These are the projects that several autonomous communities have launched, marking the way towards a multilingual digital future.

On the one hand, the Nós en Galicia Project creates oral and conversational resources in Galician with all the accents and dialectal variants to facilitate integration through tools such as GPS, voice assistants or ChatGPT. A similar purpose is that of Aina in Catalonia, which also offers an academic platform and a laboratory for developers or Vives in the Valencian Community. In the Basque Country there is also the Euskorpus project  , which aims to constitute a quality text corpus in Basque. Let's look at each of them.

Proyecto Nós, a collaborative approach to digital Galician

The project has already developed three operational tools: a multilingual neural translator, a speech recognition system that converts speech into text, and a speech synthesis application. These resources are published under open licenses, guaranteeing their free and open access for researchers, developers and companies. These are its main features:

• Promoted by: the Xunta de Galicia and the University of Santiago de Compostela.
• Main objective: to create oral and conversational resources in Galician that capture the dialectal and accent diversity of the language.
• How to participate: The project accepts voluntary contributions both by reading texts and by answering spontaneous questions.
  
  • Donate your voice in Galician: https://doagalego.nos.gal

Aina, towards an AI that understands and speaks Catalan

With a similar approach to the Nós project, Aina seeks to facilitate the integration of Catalan into artificial intelligence language models.

It is structured in two complementary aspects that maximize its impact:

• Aina Tech focuses on facilitating technology transfer to the business sector, providing the necessary tools to automatically translate websites, services and online businesses into Catalan.
• Aina Lab promotes the creation of a community of developers through initiatives such as Aina Challenge, promoting collaborative innovation in Catalan language technologies. Through this call , 22 proposals have already been selected with a total amount of 1 million to execute their projects.

The characteristics of the project are:

• Powered by: the Generalitat de Catalunya in collaboration with the Barcelona Supercomputing Center (BSC-CNS).
• Main objective: it goes beyond the creation of tools, it seeks to build an open, transparent and responsible AI infrastructure with Catalan.
• How to participate: You can add comments, improvements, and suggestions through the contact inbox: https://form.typeform.com/to/KcjhThot?typeform-source=langtech-bsc.gitbook.io.

Vives, the collaborative project for AI in Valencian

On the other hand, Vives collects voices speaking in Valencian to serve as training for AI models.

• Promoted by: the Alicante Digital Intelligence Centre (CENID).
• Objective: It seeks to create massive corpora of text and voice, encourage citizen participation in data collection, and develop specialized linguistic models in sectors such as tourism and audiovisual, guaranteeing data privacy.
• How to participate: You can donate your voice through this link: https://vives.gplsi.es/instruccions/.

Gaitu: strategic investment in the digitalisation of the Basque language

In Basque, we can highlight Gaitu,  which seeks  to collect voices speaking in Basque in order to train AI models. Its characteristics are:

• Promoted by: HiTZ, the Basque language technology centre.
• Objective: to develop a corpus in Basque to train AI models.
• How to participate: You can donate your voice in Basque here https://commonvoice.mozilla.org/eu/speak.

Benefits of Building and Preserving Multilingual Language Models

The digitization projects of the co-official languages transcend the purely technological field to become tools for digital equity and cultural preservation. Its impact is manifested in multiple dimensions:

• For citizens: these resources ensure that speakers of all ages and levels of digital competence can interact with technology in their mother tongue, removing barriers that could exclude certain groups from the digital ecosystem.
• For the business sector: the availability of open language resources makes it easier for companies and developers to create products and services in these languages without assuming the high costs traditionally associated with the development of language technologies.
• For the research fabric, these corpora constitute a fundamental basis for the advancement of research in natural language processing and speech technologies, especially relevant for languages with less presence in international digital resources.

The success of these initiatives shows that it is possible to build a digital future where linguistic diversity is not an obstacle but a strength, and where technological innovation is put at the service of the preservation and promotion of linguistic cultural heritage.