Scrutiny continues as the AI Act reaches implementation

After three years of negotiations and worldwide attention, the EU Artificial Intelligence Act entered into force 1 Aug. Attention remains high, assured Member of the European Parliament and former AI Act co-rapporteur Brando Benifei during a 25 Sept. Euractiv event. As it prepares for the hearings of Commissioners, "the European Parliament will scrutinize what is the prospect for the new European Commission, including what to do on tech," Benifei said.

In Benifei's sight, is the implementation of the AI Act, which he played a leading role in developing. Per the act itself, there is still work to do as the European Commission must develop no less than 30 guidelines, codes of practice, codes of conduct and other secondary instruments within the coming months and years. These texts will cover various key aspects of the AI Act, such as the obligations of general-purpose AI models providers, the definition of risk, technical documentation and fundamental rights impact assessments. "We need the Commission, the AI Office to finalize the guidelines regarding the prohibitions, (deliver on the) code of practice for the most powerful models," as aspect that attracts global attention, he emphasized.

As so much remains to be done, Benifei assured attendees that Parliament "has a clear understanding, shared by the Commission, of the need to be very efficient in implementing the many pieces of legislation that were adopted during the last term, as well as complementing with other initiatives around: investment, common research, strengthening the financial markets capability to sustain these tech developments and continued training."