EU authorities are weighing tighter regulation of large AI models, including Meta’s Llama 2 and OpenAI’s GPT-4. Bloomberg reports.
According to the agency, the discussion concerns additional restrictions under the forthcoming Artificial Intelligence Act. Representatives from Schengen-area member states, the European Commission, and the European Parliament want to tighten scrutiny of the leading players while avoiding overburdening startups, sources said.
They added that the talks are still at a preliminary stage.
The European Commission introduced the AI regulation bill in April 2021. It has come under criticism from civil rights advocates, who urged banning the use of tracking algorithms, including facial recognition.
In November 2022, the final version of the document appeared — it does not restrict the powers and independence of the government bodies responsible for protecting fundamental rights and freedoms. This means the bill does not prohibit the use of AI for national security purposes.
In spring 2023, EU authorities intend to require developers of generative AI to disclose any copyrighted materials used to train algorithms.
Earlier, a group of European lawmakers announced plans to draft a set of rules to govern the development of AI systems.
In May, OpenAI CEO Sam Altman urged regulation of the use and development of artificial intelligence.
CEO Coinbase Brian Armstrong, on the contrary, has spoken out against such an approach and called for decentralization of the industry.