UK MPs Say AI Safety Gaps Risk a £400bn Economic Gain
A Commons committee, citing OECD estimates, says AI could leave the UK economy up to £400 billion larger after ten years. It wants a new law and a separate regulator that can delay or prohibit unsafe training and releases.

Citing OECD estimates, a House of Commons committee says artificial intelligence could leave the UK economy up to £400 billion larger after ten years, but only if ministers close ten gaps in how the most powerful systems are tested and stopped.
The Business, Innovation, Science and Trade Committee published its report on 18 November. The document, titled Guardrails for growth: AI safety and security, is a set of recommendations, not a new law. Ministers have two months to reply.
The committee converts an OECD estimate of 0.4 to 1.3 percentage points a year in extra UK labour productivity growth into up to £38 billion of additional output a year. It says people and businesses will not adopt the technology at scale unless they trust it.
The ten gaps the MPs identify
The UK has no law governing the safety of frontier AI, meaning the most capable general-purpose models. The AI Security Institute, a government testing body set up in 2023, is not a regulator and relies on voluntary access. The report says that does not guarantee inspection before a release, or action when testers and a company disagree.
The ten gaps include no agreed definition of acceptable risk, no legal right to inspect models, no independent check of safety claims, no duty to report incidents and near misses, and no power for a public body to halt a deployment.
A new law and a separate regulator
The committee wants a legislative plan that includes an AI Safety and Security Bill, a proposed new law, in the next parliamentary session. It also wants a statutory regulator, kept separate from the Institute so the testing body can still work with firms.
Where risk cannot be adequately controlled, the government should be able to delay, restrict or prohibit a model’s training, testing or deployment in the UK. Developers of systems officially classed as high-risk would have to back up their safety claims and give independent evaluators the access needed to examine them.
The MPs also recommend a legal duty on developers to prevent harm, including after release. Some models cannot reliably be recalled once widely copied. An international agreement, including through the UK’s coming G20 presidency, should not delay rules at home. The report asks ministers to publish G20 objectives within three months.
Why it matters: The same official report that sets out a £400 billion economic gain also states that no UK regulator can currently pause a model’s training or stop a release. Until ministers accept or reject that, the testing body will still rely on companies choosing to cooperate.


