New: EU AI Regulation Agreement Redefines Tech Governance
In a watershed moment, the European Union (EU) has reached a monumental agreement on the world’s first laws to regulate artificial intelligence (AI). This historic pact, following an exhaustive 37-hour negotiation between the European Parliament and EU member states, signifies a pivotal shift in global AI governance.
Thierry Breton, the European Commissioner spearheading this regulatory framework, aptly described the accord as “historic.” Beyond AI, this groundbreaking agreement extends its reach to govern social media and search engines, casting a regulatory net that encompasses tech giants like X, TikTok, and Google.
The Marathon Negotiation
Breton revealed that a staggering 100 individuals were engaged in intense negotiations over nearly three days to finalize this momentous deal. He emphasized that the outcome was well worth the few hours of lost sleep, underscoring the historical significance of the agreement.
Key Stakeholders and Support
Carme Artigas, Spain’s secretary of state for AI, played a pivotal role in facilitating these negotiations. Reports surfaced indicating strong support from France and Germany, highlighting a collective effort to balance stringent regulation with fostering innovation, particularly among smaller companies.
Surpassing Global Competitors
This agreement places the EU at the forefront of AI regulation, outpacing global competitors such as the US, China, and the UK. The regulatory framework aims to safeguard the public from potential risks associated with AI, including threats to life—a concern echoed by many as AI technology rapidly advances.
While the details of the eventual law remain somewhat elusive, the agreement is set to take effect no earlier than 2025. The negotiation process involved intense debates, including clashes over foundation models designed for general purposes and protracted discussions on AI-driven surveillance.
Navigating AI Surveillance
One of the key points of contention was AI-driven surveillance, a technology with potential applications in policing, employment, and retail. The European Parliament successfully secured a ban on real-time surveillance and biometric technologies, including emotional recognition. However, Breton outlined three exceptions, permitting usage in cases of unexpected terrorist threats, victim searches, and the prosecution of serious crimes.
Safeguarding Against Abuse
MEP Brando Benefei, co-leading the parliament’s negotiating team, emphasized the importance of ensuring independent authorities’ permission for “predictive policing.” This move seeks to prevent potential abuses by law enforcement and upholds the principle of presumption of innocence in crime.
The foundation of the agreement adopts a risk-based tiered system, where the highest level of regulation applies to machines posing the greatest risks to health, safety, and human rights. Notably, the definition of the highest risk category now hinges on the number of computer transactions needed to train the machine, known as “floating point operations per second” (Flops). Sources suggest that only one model, GPT4, currently falls into this new definition.
As the EU sets in place real regulation for AI, it positions itself as a trailblazer in guiding the development and evolution of this transformative technology in a human-centric direction. Reflecting on past mistakes, the EU aims to avoid the pitfalls of unregulated tech growth, taking a comprehensive and proactive stance.
Anu Bradford, a Columbia Law School professor specializing in the EU and digital regulation, emphasizes that strong and comprehensive regulation from the EU could set a powerful example for governments worldwide. While other countries may not replicate every provision, they are likely to emulate many aspects, creating a global ripple effect.
AI companies adhering to the EU’s rules are expected to extend similar obligations to markets outside the continent. This move is not just a matter of efficiency but also underscores the potential influence of EU regulations on a global scale.
In conclusion, the EU AI regulation agreement is a historic landmark moment in the evolving landscape of technology governance. As the world watches and adapts to these transformative regulations, the EU stands as a beacon, leading the way into an era of responsible and human-centric AI development.