A UK parliamentary committee wants a new watchdog able to halt high-risk AI training and testing, block deployments or suspend services when risks cannot be contained. Its proposals would give independent oversight real teeth, but they are recommendations, not law.
Watch Desk analysis
What happened
The Business, Innovation, Science and Trade Committee is calling for an AI Safety and Security Bill in the next parliamentary session. The Register reports that the committee’s report, Guardrails for Growth: Making AI Safe, proposes a separate regulator with enforcement powers, while the existing AI Security Institute continues testing models.
The proposed rules would require developers of designated high-risk systems to support their safety claims with evidence and give independent researchers access to examine them. Significant changes to a model’s capabilities or safeguards would trigger further scrutiny. MPs also want compulsory reporting of incidents and near misses, and stronger protections for whistleblowers.
Why it matters
The committee says the UK’s AI Security Institute currently relies on voluntary cooperation and cannot stop a company releasing a model that raises safety concerns. A regulator with intervention powers would change who gets the final say when developers’ assurances and independent scrutiny do not line up.
The committee’s recommendations are not binding, and the Government has not committed to the proposed legislation. That leaves a substantial gap between MPs asking for a brake and ministers deciding whether to install one.
Our read
This is a consequential proposal, not a new rule for AI companies to follow today. The practical change would be a shift from voluntary access and company-set safety assurances towards mandatory evidence, independent examination and the power to intervene. That is a much more serious arrangement than asking the industry to mind the machinery itself.
What to watch
- Whether ministers include an AI Safety and Security Bill in the next parliamentary session.
- How any law would define high-risk systems and the changes that trigger further scrutiny.
- Whether the proposed regulator would have enough independence and capacity to use its powers.
Discussion spark: Should an independent regulator be able to halt high-risk AI development before deployment, or should that power remain with ministers and the companies building the models?
Sources and evidence
- UK MPs want AI watchdog with power to pull the plug on dangerous models (18 November 2026, 09:30 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.