UK Rights Committee Calls for AI Law and Regulator With Enforcement Powers

People harmed by an AI system can face a confusing path to redress in the UK, Parliament’s Joint Committee on Human Rights warns. Its answer is a new law and an independent regulator with authority to oversee AI risks from development through deployment.

The cross-party committee of MPs and peers said current laws and regulations are patchy, confused and uncoordinated. It recommended a statutory body that could set policy, monitor performance and enforce rules—a substantial change from a system where no single institution coordinates AI regulation.

The proposed framework would not apply identical duties to every system. The committee recommended lighter requirements for lower-risk uses and tougher obligations for higher-risk AI systems and models, alongside mandatory transparency throughout the AI lifecycle.

The protections the committee wants

• A bill covering protections throughout the AI supply chain and lifecycle.
• Stronger duties for higher-risk systems and models, alongside lighter rules for lower-risk uses.
• Prohibitions on AI uses judged incompatible with human rights.

The committee tied that structure to harms that are already understandable outside technical policy circles: public face-scanning, explicit deepfakes and workers being flagged for discipline without reasonable cause. It also warned that discriminatory outcomes can stem from biased training data as well as how systems are built or deployed.

The UK already has the AI Security Institute, which tests advanced models for safety. But it cannot compel developers to submit systems for scrutiny. The committee’s proposed regulator would therefore address a different institutional problem: not only whether advanced models are tested, but who can require oversight and respond to harms across the wider AI lifecycle.

The report does not itself change the law. The Labour government said after taking office in 2024 that it would seek legislation imposing requirements on developers of the most powerful models, but had not introduced that legislation when the committee reported. The government has separately said it is exploring targeted interventions for significant AI-related national-security risks.

The committee has set out broad duties for the regulator, but the government has not introduced legislation to enact them. Whether ministers adopt the proposal—and how they implement it—remains unresolved.