Legal implications of open data and re-use of public sector information for ChatGPT

The emergence of artificial intelligence (AI), and ChatGPT in particular, has become one of the main topics of debate in recent months. This tool has even eclipsed other emerging technologies that had gained prominence in a wide range of fields (legal, economic, social and cultural).  This is the case, for example, of web 3.0, the metaverse, decentralised digital identity or NFTs and, in particular, cryptocurrencies.

There is an unquestionable direct relationship between this type of technology and the need for sufficient and appropriate data, and it is precisely this last qualitative dimension that justifies why open data is called upon to play a particularly important role. Although, at least for the time being, it is not possible to know how much open data provided by public sector entities is used by ChatGPT to train its model, there is no doubt that open data is a key to improving their performance.

Regulation on the use of data by AI

From a legal point of view, AI is arousing particular interest in terms of the guarantees that must be respected when it comes to its practical application. Thus, various initiatives are being promoted that seek to specifically regulate the conditions for its use, among which the proposal being processed by the European Union stands out, where data are the object of special attention.

At the state level, Law 15/2022, of 12 July, on equal treatment and non-discrimination, was approved a few months ago. This regulation requires public administrations to promote the implementation of mechanisms that include guarantees regarding the minimisation of bias, transparency and accountability, specifically with regard to the data used to train the algorithms used for decision-making.

There is a growing interest on the part of the autonomous communities in regulating the use of data by AI systems, in some cases reinforcing guarantees regarding transparency. Also, at the municipal level, protocols are being promoted for the implementation of AI in municipal services in which the guarantees applicable to the data, particularly from the perspective of their quality, are conceived as a priority requirement.

The possible collision with other rights and legal interests: the protection of personal data

Beyond regulatory initiatives, the use of data in this context has been the subject of particular attention as regards the legal conditions under which it is admissible. Thus, it may be the case that the data to be used are protected by third party rights that prevent - or at least hinder - their processing, such as intellectual property or, in particular, the protection of personal data. This concern is one of the main motivations for the European Union to promote the Data Governance Regulation, a regulation that proposes technical and organisational solutions that attempt to make the re-use of information compatible with respect for these legal rights.

Precisely, the possible collision with the right to the protection of personal data has motivated the main measures that have been adopted in Europe regarding the use of ChatGPT. In this regard, the Garante per la Protezione dei Dati Personali has ordered a precautionary measure to limit the processing of Italian citizens' data, the Spanish Data Protection Agency has initiated ex officio inspections of OpenAI as data controller and, with a supranational scope, the European Data Protection Supervisor (EDPB) has created a specific working group.

The impact of the regulation on open data and re-use

The Spanish regulation on open data and re-use of public sector information establishes some provisions that must be taken into account by IA systems. Thus, in general, re-use will be admissible if the data has been published without conditions or, in the event that conditions are set, when they comply with those established through licences or other legal instruments; although, when they are defined, the conditions must be objective, proportionate, non-discriminatory and justified by a public interest objective.

As regards the conditions for re-use of information provided by public sector bodies, the processing of such information is only allowed if the content is not altered and its meaning is not distorted, and the source of the data and the date of its most recent update must be mentioned.

On the other hand, high-value datasets are of particular interest for these AI systems characterised by the intense re-use of third-party content given the massive nature of the data processing they carry out and the immediacy of the requests for information made by users. Specifically, the conditions established by law for the provision of these high-value datasets by public bodies mean that there are very few limitations and also that their re-use is greatly facilitated by the fact that the data must be freely available, be susceptible to automated processing, be provided through APIs and be provided in the form of mass downloading, where appropriate.

In short, considering the particularities of this technology and, therefore, the very unique circumstances in which the data are processed, it seems appropriate that the licences and, in general, the conditions under which public entities allow their re-use be reviewed and, where appropriate, updated to meet the legal challenges that are beginning to arise.

Content prepared by Julián Valero, Professor at the University of Murcia and Coordinator of the "Innovation, Law and Technology" Research Group (iDerTec).

The contents and points of view reflected in this publication are the sole responsibility of the author.