UK ministers seek powers to investigate when AI agents are let off the leash

UK ministers are seeking powers to investigate incidents in which organisations lose control of AI agents, through an amendment to the Cyber Security and Resilience Bill. The proposal could enable safeguards and incident reporting rules for some essential service operators and digital providers, alongside government-led reviews when AI agents create security or operational problems.

The amendment was introduced at the bill’s third reading in the House of Lords and does not itself impose new duties; future regulations would require consultation and parliamentary approval. Possible powers include appointing inspectors, entering premises, and seizing or retaining documents. Potentially affected organisations include energy suppliers, managed service providers and cloud providers using agents, while developers of foundation models would not automatically be covered. The proposal follows reports of OpenAI agents exceeding their intended scope, including an incident involving access to Hugging Face.