EU countries reach agreement on AI

(ANSA) - ROME, FEB 5 - (by Alessandra Briganti) The last obstacle to the final approval of the EU's AI Act has been cleared. The ambassadors of the 27 member states have unanimously approved the agreement reached last December by the EU institutions on the ground-breaking law laying down the rules for the development, marketing and use of AI systems in Europe.
    The outcome was by no means a foregone conclusion.
    The text, the result of three years of arduous work and the first of its kind in the world, appeared to teeter after the agreement, with France leading the discontent. For Paris the imperative is that Europe does not fall behind in the race to develop the new technology, as happened with the internet revolution and social platforms. France, supported by Italy and Germany, took issue in particular with the rules on founding models, such as ChatGPT, for which the proposed law envisages a multi-level approach, with horizontal transparency rules for all models and stricter obligations for those presenting systemic risk.
    However, now even the last doubts have been dispelled. Or rather, postponed to the critical implementation phase. Doubts that for Paris concern not only innovation, but also the balance between transparency and the protection of trade secrets.
    Austria, on the other hand, registered its concerns about data protection and the rules that prohibit, albeit with exceptions, certain AI practices, such as biometric remote identification.
    Despite the qualms, the European Commissioner for the Internal Market, Thierry Breton, toasted the go-ahead from the 27 member states, describing it as "a historic first" that recognises "the balance achieved by the negotiators between innovation and security".
    European Commission Vice President Margrethe Vestager said on X that the case of Taylor Swift, an illustrious victim of AI, "says everything about the damage AI can cause if misused, the responsibility of platforms and why it is so important to enforce technology regulation".
    Brando Benifei, PD MEP and rapporteur on the AI regulation, hailed the "good day for the EU" and also stressed the urgent need to "support the early application of the rules for countering disinformation and deepfakes to protect our democracies in the most important election year for Europe and the world".
    Rome also expressed its satisfaction, with cabinet undersecretary responsible for technological innovation, Alessio Butti, speaking of a "historic moment for the EU" and emphasising "the decisive contribution of Italy and the Meloni government" to the agreement on an issue that is central to Rome.
    AI will in fact be one of the topics of Italy's duty presidency of the G7 this year.
    "Our aim was to ensure that all AI applications, including cutting-edge generative models, operate within a system of rules that is both simple and rigorous, capable of protecting citizens' rights and promoting responsible innovation," continued Butti.
    The ball is now in the European Parliament's court, where the Single Market and Civil Liberties committees are scheduled to vote on the text on February 13 before it goes to the assembly for the final go-ahead, presumably in the April session. The regulation will then finally have to be ratified by the EU Council.
    (Photo: Cabinet Undersecretary for technological innovation, Alessio Butti.) (ANSA).