EU gains power to regulate AI models

The European Union is set to gain new powers to police large general purpose AI models, with the ability to demand information, conduct safety evaluations, and impose fines of up to 3% of annual total turnover. This move places the EU at the forefront of global AI safety regulation, with the new mandate taking effect on August 2.

The EU’s AI Act targets providers of the largest general purpose AI models, including companies like OpenAI, Anthropic, Meta, Alphabet, xAI, Alibaba, ByteDance, and Z.ai. France’s Mistral is the only European model provider expected to fall under the law.

Enforcement Era Begins

The August 2 launch has been years in the making, with the EU simplifying its digital rulebook and delaying rules governing high-risk AI systems. The rules require companies to perform state-of-the-art model evaluations, assess and mitigate systemic risk, report serious security incidents, and ensure appropriate cybersecurity protection.

Risto Uuk, head of European policy and research at the Future of Life Institute, notes that enforcement depends on the willingness at a political level to enforce the law and take it seriously. “A lot of stakeholders are watching and hoping that they will be strong from day one,” he said.

The AI office has built much of the infrastructure needed to enforce the rules, including publishing the Code of Practice, issuing guidelines for model providers, and establishing an AI Act Advisory Forum. However, the office may not have enough personnel to enforce the law, with a report estimating that 160 staff will be needed by 2030, compared to the current 40.

Jimmy Farrell, EU AI policy lead at Pour Demain, expects the AI Office’s first enforcements to involve requests for information and structured dialogue with model providers, rather than blockbuster fines.

The AI Office’s success will depend on avoiding politicization, prioritizing the largest safety risks, and being adaptable to rapid technological changes. With international support from the US and China, Europe may not feel entirely alone in its efforts to regulate frontier AI models.

As the EU enters this new phase, it will be closely watched to see how it balances enforcement with the need to attract talent and investment in the AI sector. The AI Office’s ability to respond to emerging needs and demonstrate the effectiveness of its regulations will be essential to its success. European developers, including those in the US and China, will be subject to scrutiny, with the Future of Life’s summer 2026 AI Safety Index giving failing grades to xAI and Mistral.

Enforcement Challenges Ahead

The AI Office’s limited personnel may hinder its ability to effectively enforce the rules. With only around 40 staff as of July, the office may struggle to keep up with the demands of policing large general purpose AI models.