World’s Most Extensive AI Rules Approved in EU

Summary

The EU's new AI act has been passed, raising concerns among tech companies about overregulation and competitiveness with US counterparts. The act focuses on a risk-based approach, banning high-risk AI applications like emotion recognition and social scoring. Companies like Mistral, now partnered with Microsoft, must prove compliance with safety and copyright laws to the new EU AI office, which can demand more data and even ban applications. Despite initial lobbying against strict controls, some tech companies are now embracing voluntary commitments to show they take regulation seriously.

Takeaways

• 🎉 The passing of the act is being celebrated in Brussels, marking the end of nearly 40 hours of negotiations and three months of deliberation.
• ⚠️ Tech companies, including European startups, have expressed concerns about overregulation and the potential for the European continent to fall behind its U.S. and Asian counterparts.
• 🚫 The act prohibits AI systems from being used for emotion recognition in workplaces or schools and for social scoring, which assigns citizens a score based on their behavior.
• 🛑 High-risk AI applications, such as those used in migration applications or job applications, will require companies to perform additional checks to prove their safety to regulators.
• 🔍 The new EU AI office in Brussels will act similarly to a police force, with the power to request more data on AI training and even ban poorly performing applications.
• 🤝 Gillian Misrule's partnership with Microsoft highlights a shift in tech companies' stance, from lobbying against strict controls to forming strategic partnerships.
• 🌐 The EU's approach to AI regulation may influence how the rest of the world adopts AI legislation, as seen with tech companies pushing for regulation while lobbying against strict controls.
• 🤔 Tech companies are beginning to realize that their lobbying efforts may be seen as self-serving, especially when they partner with large corporations shortly after advocating against certain regulations.
• 📜 Companies like Mistral will need to comply with the new regulations, which focus on the use of technology and not the technology itself, emphasizing a risk-based approach.
• 🏛️ The EU's legislation includes additional controls for general-purpose or generative AI, which is a departure from explicitly regulating the technology itself.
• 📈 Tech companies are signing on for voluntary commitments to demonstrate to governments that they are taking the new regulations seriously and are committed to compliance.

Q & A

What is the current sentiment in Brussels regarding the new act?

The current sentiment in Brussels is of celebration as the parliament has signed off on the act, which has been the result of nearly 40 hours of negotiations and months of anticipation.

What are the main concerns of tech companies about the new act?

Tech companies are concerned about overregulation, fearing that the European continent will fall behind its U.S. and other international counterparts due to more stringent regulations.

How do European tech companies and startups view the new regulations?

European tech companies and startups are worried that overregulation will hinder their ability to compete with U.S. hyperscalers, as they believe strict controls could stifle innovation and growth.

What is Mistral's position on the new act?

Mistral, a company based in Europe building large language models, finds the new act to be problematic and is seeking clarity on how to comply with the regulations.

What is the EU's approach to regulating AI?

The EU's approach is risk-based, focusing on the use of technology rather than the technology itself, aiming to ban the worst possible uses of AI such as emotion recognition in workplaces or social scoring.

What are some high-risk situations mentioned in the act for AI systems?

High-risk situations include AI systems being used for migration applications or starting job applications, where companies like OpenAI or Mistral will need to perform more checks to prove safety to regulators.

What exception did companies lobby hard against?

Companies lobbied hard against additional controls on general-purpose or generative AI, which are not based on the specific use of AI but rather on the technology itself.

What will companies need to prove to regulators regarding their AI systems?

Companies will need to prove their energy consumption and compliance with copyright laws to regulators, and may be subject to scrutiny by the new AI office based in Brussels.

How does the new AI office in Brussels function?

The new AI office in Brussels operates almost like a police force, with the power to request more data on how companies have trained their large language models and even ban an application if it's performing poorly.

How has the partnership between Gillian Misrule and Microsoft affected EU lawmakers?

The partnership has left a bad taste in the mouths of many lawmakers, as it seems contradictory to the lobbying efforts against strict regulations, raising questions about the self-serving nature of tech companies' actions.

What is the potential global impact of the EU's AI legislation?

The EU's AI legislation could influence how the rest of the world adopts AI regulations, as it has seen tech companies pushing for regulation while simultaneously lobbying against strict controls, indicating a complex dynamic between regulation and competition.