In discussions leading up to the final meetings scheduled for December 6, “platform models” or generative AI have emerged as a major hurdle, the anonymous source said.
Platform models like the one built by OpenAI by Microsoft are AI systems trained on large datasets, with the ability to learn from new data to perform a variety of tasks.
After two years of negotiations, the bill was approved by the European Parliament in June. However, the content of the draft AI regulation now needs to be agreed through meetings between representatives of the European Parliament, the Council and the European Commission.
While some experts and lawmakers have proposed a tiered approach to regulating platform models, such as a platform user base of 45 million or more, others have argued that smaller models could be equally at risk.
But the biggest challenge to reaching an agreement comes from France, Germany and Italy — countries that favor allowing AI model makers to self-regulate rather than having rigid rules.
European MEPs, EU Commissioner Thierry Breton and many AI researchers have opposed leaving AI companies to “self-regulate”, as this “likely fails to meet the standards needed for the safety of the platform model”.
On the other hand, businesses don’t want to be held to rigid standards for the technology they use to create their products. France-based AI company Mistral and Germany’s Aleph Alpha have criticized the hierarchical management approach.
Reuters sources said Mistral favored hard rules on products, not the technology used inside.
Divided and Ambiguous
“While stakeholders are doing their best to keep negotiations on track, the growing regulatory uncertainty is not in the best interests of European industries,” said Kirsten Rulf, partner and associate director at Boston Consulting Group. “European businesses want to plan for the year ahead, and many would like to see some certainty around the EU AI Act, which comes into force in 2024.”
Other pending issues in the negotiations include the definition of AI, fundamental rights impact assessments, law enforcement exceptions and national security exceptions.
Lawmakers are also divided over law enforcement's use of AI systems to biometrically identify individuals in public spaces.
Spain, which holds the rotating EU presidency until the end of the year, has been trying to speed up the process of reaching an agreement.
If a deal is not struck by December, the next presidency, Belgium, will have just months to get it done before the whole thing is likely to be shelved until after the European elections.
ChatGPT's year of 'storming the world'
OpenAI Looks to Bring ChatGPT into the Classroom
EU reaches agreement on Smart Device Protection Act
Source
Comment (0)