EU officials sign ‘historic’ AI regulatory deal
EU Parliament and Council negotiators reached a provisional agreement on rules governing the use of artificial intelligence on Friday, December 8.
The agreement includes government use of AI for biometric surveillance, how AI systems like ChatGPT will be regulated, and transparency rules they must follow before entering the market. This includes sharing technical documentation, EU copyright compliance and training content summaries.
The EU wants to be the first supranational authority to draft laws on AI, setting out how AI can be used for benefit while protecting against risks. The deal was concluded after nearly 24 hours of discussions on December 8 and 15 hours of negotiations thereafter.
The agreement requires AI models with significant impact and systemic risk to assess and address these risks, conduct adversarial testing of system resilience, report incidents to the European Commission, ensure cybersecurity, and disclose energy efficiency. It is stipulated that it should be done.
“Proper implementation is essential, and Congress will pay close attention to supporting new business ideas through sandboxes and effective rules for the strongest models.”
After the deal was concluded, European Commissioner Thierry Breton said: Posted “Historic!” said X, formerly known as Twitter. #AIAct is more than a rulebook. This is a stepping stone for EU startups and researchers to lead the global AI race. “The best is yet to come!”
historic!
The EU has become the first continent to set clear rules for the use of AI.
that much #AI Act This book is more than just a rule book. It is a stepping stone for EU startups and researchers to lead the global AI competition.
The best is yet to come! pic.twitter.com/W9rths31MU
— Thierry Breton (@ThierryBreton) December 8, 2023
According to the agreement, any general purpose artificial intelligence (GPAI) that poses a risk must follow the code. Governments can only use real-time biometric surveillance in certain cases, such as certain crimes or serious threats in public places.
Related: AI regulation is in focus globally as the EU approaches regulatory negotiations.
The deal bans biometric systems that manipulate cognitive behavior, scrape facial images from the internet or CCTV footage, social score, and infer personal information such as beliefs and tendencies. Consumers have the right to complain and receive an explanation.
Fines for violations range from $8.1 million (€7.5 million), or 1.5% of turnover, to $37.7 million (€35 million), or 7% of global turnover, depending on the extent of the infringement and the size of the company.
According to a statement from the European Parliament, what was agreed must now be formally adopted by the Parliament and Council before it becomes EU law. Parliament’s Internal Market and Civil Liberties Committee is expected to vote on the agreement at an upcoming meeting.
magazine: Real-life AI use cases in cryptocurrency: Cryptocurrency-based AI market and AI financial analysis
