MPs and peers want a new AI law and a statutory watchdog — the government says Britain already leads the world

A cross-party committee of MPs and peers has told the government that Britain's laws are not built for what artificial intelligence can already do to people's rights — and that only a new Act of Parliament, with a regulator behind it, will close the gap.

The Joint Committee on Human Rights published its report on Monday. Its central finding is blunt: the country has no framework equal to the technology it is trying to govern.

"Nowhere in the world, including the UK, has a current legislative and regulatory approach to AI that is fit for purpose," said the committee's chair, the Labour MP Alex Sobel. AI, he added, "is moving with such speed and complexity that its impact is hard to accurately predict."

What the committee wants

The report does not ask for a code of practice or a set of principles. It asks for law. Three things sit at the centre of it:

• A new AI Bill written specifically around human-rights risk, rather than bolted onto existing sector rules.
• A single, independent AI oversight body on a statutory basis — one regulator with legal standing, not a patchwork of existing watchdogs each covering a corner.
• A risk-tiered regime, placing heavier legal duties on higher-risk systems and models, with those duties applying across the whole AI lifecycle: the people who design a system, the people who amend it, and the people who use it.

The committee also wants some uses banned outright as incompatible with human rights — among them subliminal manipulation techniques and inappropriate use of profiling and biometric data.

The harms it names

The report is specific about what it is worried about, which is part of why it lands harder than the usual call for "guardrails". It points to the creation of sexualised images of women and girls, facial scanning carried out without consent, bias against particular groups baked in from internet-scraped training data, and the manufacture and spread of misinformation.

These are not future risks. Each is a documented feature of systems already in use, which is the committee's point: the law is behind the deployment, not behind the research.

The government's answer

Ministers did not accept the framing. A government spokesperson pointed to work already under way: "We're leading global efforts to test and understand AI, with the best-resourced security institute in the world, that will help us develop the standards and safeguards necessary to keep people safe."

That is the fault line. The committee's case is that testing and understanding are not the same as enforceable rights, and that an institute which studies models is not a regulator that can compel their makers. The government's case is that Britain is already ahead and that binding law risks slowing the thing it is trying to lead on.

Why it matters here

The report lands days after the people building frontier AI — Dario Amodei, Sam Altman and Elon Musk — publicly agreed the pace should slow, and a day after the US president rejected those same warnings as overblown. Westminster is now the venue asking the harder, less headline-friendly question: not whether AI should slow down, but who is legally on the hook when it harms someone.

A committee report is a recommendation, not a law. The government is under no obligation to act on it, and its response suggests it is in no hurry to. But it puts a concrete proposal — a bill, a statutory regulator, a risk regime — on the table in the UK, and names, in plain terms, the people already being harmed while the framework is argued over.