With the Commission for the Guarantee of the Dret d'Accés to Public Information (GAIP) As the organizer of the meeting, the transparency guarantee bodies of the entire country have met for two days, the 4 and 5 June, to discuss, share experiences and continue strengthening institutional collaboration to promote transparency and the right of access to public information.

In her speech, the president of the CTBG recalled that Law 19/2013 on transparency, access to public information and good governance was designed to guarantee access to documents, but today it must face requests aimed at algorithms, source code, automated decision models and artificial intelligence systems used by public administrations in decision-making that directly affect citizens.

Campos considers it essential to tackle two inalienable fronts. On the one hand, take advantage of the potential of AI as a working tool for the bodies that guarantee transparency, without this in any case replacing the work of the people who instruct the files, but as an aid to automate tasks such as the classification of claims, the continuous evaluation of transparency portals or configuring trained resolution assistants on the doctrine of transparent bodies. “Not to replace the instructor, but so that no relevant precedent remains unreviewed and doctrinal coherence is real, not aspirational,” he said.

On the other hand, in his opinion, it is essential to define the scope that AI transparency should have. “When the Administration uses an algorithm to make decisions about citizens, to award aid or support the decision-making process, the right of access does not disappear due to the fact that the information is in source code, in an algorithm or in a computer application,” he said, and an integrating reading of the LTAIBG must be made with the European Union Artificial Intelligence Regulation (2024/1689), which has been fully applied since August 2026, and with Article 22 of the General Data Protection Regulation.

“It is necessary to clarify whether the technical documentation required by the European AI Regulation for high-risk systems – including those used in the management of social benefits, public employment, the administration of justice and migration management – constitutes public information accessible under the LTAIBG.” “It’s time to review the damage test in a technological key. Restricting access to the code of a system that makes decisions about citizens may, in itself, be the greatest possible harm to the public interest,” he said.

In this regard, he said that the Council of Transparency and Good Governance will work on the elaboration of an interpretative criterion that distinguishes source code, functional documentation and explanation of results, with a differentiated damage test for each level. He also proposed a pilot project to apply AI to the management of transparency and the collaboration of all guarantor bodies, the state and the autonomous, to work in a network and build a common doctrine on algorithmic transparency.

COTAI was established in May 2025 as an entity, without legal personality, of horizontal collaboration composed of the Council and the autonomous supervisory bodies in the field of transparency, with an annual rotating presidency.

Presides in 2026 the COTAI la GAIP, whose president, Iolanda Pineda Balló, highlighted at the opening of the conference in Barcelona the opportunity that this meeting represents to “contrast points of view, discuss the challenges we face and build common responses or agree on discrepancies”. Pineda recalled the key role played by the councils and transparency commissioners to “strengthen the democratic quality and the confidence of the citizens in the public powers.”

During the meeting, which concludes today, the participants highlighted, among other issues, the need to have mechanisms to strengthen compliance with their resolutions, such as coercive and sanction instruments, and their concern about the progressive increase in claims, without greater human resources being provided to guarantee bodies, which are scarce and limited already in the current context.