Sources, EU Commission to present AI Model Code on Thursday

As far as we know, the European Commission to present the Code of Ethics on Thursday 10 July best practices on artificial intelligence models for general purpose AI (GPAI), such as OpenAI's GPT-4 and Gemini Google. The Code, whose final draft was expected to May, will have to detail the provisions of the AI ​​Act that will will apply, starting from August 2, to model suppliers Gpai, especially those with systemic risks. The third draft of the code had raised protests from the insiders, including the negotiators in Parliament themselves European and the EU Council, according to which the text was too biased in favor of Big Tech. Parliamentary sources they report an improvement of the Code which is being finalized compared to the latest draft, especially in relation to the assessment and mitigation of risks to human rights fundamental. As for the implementation of the code, the spokesperson of the European Commission Thomas Regnier explained that "the Committee European Commission for AI is discussing the timing", in particular "the possibility of evaluating the end of 2025". "This - he observed - it would be well ahead of the implementation deadlines of the AI ​​Act: 2026 for new models and 2027 for existing models existing". In the event that a company does not sign the code of good practices, the AI ​​Act deadlines will continue to be valid, but will not benefit from the legal certainty offered by the Code". "The powers of the Commission for the application of these rules will only take effect from 2 August 2026", the spokesperson specified, explaining that this applies to new models released on the market after August 2, 2025. The existing models, which are the majority, must comply by August 2, 2027.