EU is Close to an Agreement on the Regulation of ChatGPT and Other AI-Based Systems
EU | December 7, 2023, Thursday // 11:37| viewsThe European Union is close to a deal on what could become the largest and most far-reaching regulation of artificial intelligence in the Western world, Bloomberg reports.
Negotiators agreed early today on a set of controls for generative artificial intelligence tools like OpenAI Inc.'s ChatGPT. and Google's Bard, the language model capable of producing content on command, sources familiar with the ongoing discussions told Bloomberg.
Meanwhile, the Spanish news agency EFE cited its sources as saying that European institutions have already reached an agreement on the norms that will regulate systems based on artificial intelligence.
Delegates from the European Commission, the European Parliament and the 27 member states reached the compromise at a meeting that began Wednesday afternoon and lasted for hours, bringing the group closer to formal agreement on a broader piece of legislation known as the Artificial Intelligence (AI) Act.
The deal marks a key step in clearing up AI policy, which – in the absence of any meaningful action by the US Congress – will set the tone for regulation of generative AI tools like OpenAI's ChatGPT and Google's Bard in the developed world.
European politicians have been working for months to finalize the texts of the "AI Law" and pass it before the European elections in June next year, which will lead to the formation of an entirely new European Parliament and European Commission, but which could impose more changes and stop efforts in this direction.
The wide-ranging discussions late last night and early this morning underscored how contentious the debate over AI regulation has become, dividing world leaders and tech executives. The EU — like others, including the US and the UK — is struggling to balance the need to protect its own AI startups, such as France's Mistral AI and Germany's Aleph Alpha, against potential societal risks.
That has proved a key sticking point in the talks, with some countries, including France and Germany, opposing the rules, which they say would unnecessarily burden local companies.
European politicians have proposed a plan that would require developers of the types of AI models that underpin tools like ChatGPT to maintain information about how their models are trained, summarize the copyrighted material used and label the AI-generated content.
Systems that pose "systemic risks" will have to work with the European Commission through an industry code of conduct. They will also have to monitor and report any incidents from the AI models.
Back