NVDA 145.98 ▼3.30%GOOGL 272.66 ▼3.46%MSFT 498.30 ▼1.78%AMD 489.19 ▼7.72%INTC 55.72 ▲1.92%TSMC 352.09 ▼2.90%AMZN 257.37 ▼3.87%META 596.22 ▼3.27%AAPL 349.21 ▼0.92%PLTR 153.68 ▼1.91%
Markets at last close

Policy

UK MPs seek watchdog powers to halt unsafe AI

·1 min read

UK MPs are calling for an AI Safety and Security Bill that would empower a watchdog to halt high-risk training and testing, block deployments and suspend services when risks cannot be contained. Parliament’s Business, Innovation, Science and Trade Committee wants intervention during development, warning that some models cannot reliably be recalled after release.

OpenAI, Google DeepMind, Anthropic and Meta told the committee they would not release models they considered unsafe, but their safety standards differ. UK law generally does not require independent assessments before release. The proposed framework would require developers of designated high-risk systems to substantiate safety claims and provide access for independent researchers, with further scrutiny following significant changes to capabilities or safeguards.

The AI Security Institute would retain its testing role, while a separate regulator would enforce the rules. Its current work relies on voluntary cooperation, and it cannot stop releases. MPs also want compulsory reporting of incidents and near misses, alongside stronger legal protections for staff whose safety concerns may be suppressed by restrictive contracts, commercial pressures or financial disincentives.

The committee argues stronger safeguards could encourage business adoption and support a market for independent testing. It wants legislation in the next parliamentary session and international coordination during Britain’s G20 presidency in 2027. The recommendations are not binding, and ministers have yet to commit to the proposed law.

Originally reported by theregister.comRead the source →
Related coverage