EU negotiations on AI Act go into extra time

(ANSA) - ROME, DEC 8 - (by Valentina Brini) The date of December 7 had been marked in red on the calendar, but 22 hours of negotiations were not sufficient for a final result.
    The European Union's AI Act, the world's first law on artificial intelligence, is one step closer to seeing the light of day, but extra time will be needed to conclude the historic agreement.
    In the new chapter of the eternal challenge between man and technology, the positions of national governments and the European Parliament still need to come closer to clearly delimit the scope of AI with the aim of putting the protection of fundamental rights centre stage.
    In the common European approach, these cannot be sacrificed even on the altar of an economic race with the United States and China that for the EU is worth 20 billion euro in investments per year until 2030.
    The last remaining sticking points concern facial recognition - rejected in its entirety by MEPs and instead admitted for security reasons by national governments - and the regulation of ChatGpt, Bard and other AI systems based on foundational models.
    The afternoon and night of negotiations in Brussels were followed by a dawn still full of hope.
    Then, after lunch, came the feeling of stalemate and the awareness that it would be necessary to resume work the following day in order to clarify ideas, iron out the last remaining differences and seal the agreement.
    In the long negotiations, EU Commissioner Thierry Breton, the "father" of the legislative proposal dated April 2021, said "a lot of progress" had been made - enough not to overturn the negotiating table and to nourish optimism for a happy ending.
    This will inevitably come via the path of exceptions to be granted - or not - on mass surveillance systems.
    The 27 EU member states have rejected the complete ban on a Big Brother called for by MEPs.
    The two positions have been irreconcilable since the dawn of the first Brussels proposal - which did provide for exceptions - and even in the final hours of talks the sides are struggling to find a compromise, with Madrid's EU presidency supporting exceptions for public safety, the fight against terrorism and the search for missing children or criminals.
    There is a similar contrast in relation to generative artificial intelligence systems, with MEPs calling for ad hoc supervision and more transparency on the algorithms and on the huge databases at their core and member states trying to avoid excessive regulation in the fear that it could nip emerging champions such as Aleph Alpha in Germany and Mistal AI in France in the bud, making their development costs prohibitive.
    Further negotiations will be needed to reach a final mediation agreement. (ANSA).