Neutral

MPs Demand New AI Safety Law and Regulator with Powers to Block Dangerous Systems

A Commons committee calls for an AI Safety and Security Bill and independent regulator empowered to halt high-risk AI systems, identifying ten gaps in UK safeguards.

By Staff member • Published 18 November 2026 at 12:45 • 3 min read
Share article X Facebook LinkedIn WhatsApp Email

MPs have called for new legislation giving an independent regulator the power to halt the development and deployment of dangerous artificial intelligence systems, warning that Britain has significant gaps in its safeguards against emerging AI risks.

The House of Commons Business, Innovation, Science and Trade Committee has urged the Government to introduce an AI Safety and Security Bill during the next parliamentary session. Its report, Guardrails for Growth: AI Safety and Security, published on 18 November 2026, identifies ten critical weaknesses in Britain’s approach to advanced AI safety.

The committee says stronger oversight is needed to protect the public while enabling the UK to benefit from the economic opportunities offered by artificial intelligence.

Ten gaps in AI safety protections

Following evidence from researchers, former industry employees and representatives of OpenAI, Google DeepMind, Anthropic and Meta, MPs identified weaknesses including unclear thresholds for unacceptable harm, limited independent verification of safety claims, inadequate incident reporting, insufficient whistleblower protections and uncertain legal liability.

The committee warns that the existing system relies heavily on voluntary cooperation between developers and the AI Security Institute (AISI). AISI conducts technical research and evaluations but is not a regulator with powers to approve or prohibit deployment.

Regulator could pause training and block deployment

The report recommends a statutory AI regulator separate from AISI, with secure multi-year funding and powers to obtain access to advanced models and relevant technical information. Developers of designated high-risk systems should have to substantiate safety claims and permit independent scrutiny.

Where risks cannot be adequately controlled, MPs want powers, in defined circumstances, to require safeguards, pause training or testing, delay or prohibit deployment in the UK, and restrict or suspend unsafe services. Ministers accountable to Parliament would remain responsible for determining acceptable risk thresholds.

Warning over models that cannot be recalled

MPs highlight the difficulty of controlling open-weight AI models once their underlying weights have been downloaded, copied or modified by third parties. They recommend independent scrutiny before high-risk releases and emergency plans involving developers, hosting services, regulators and international partners.

The committee also wants authorities to have effective measures against harmful misuse, recognising that a widely distributed model may be impossible to recall completely.

Mandatory reporting and whistleblower protection

The proposed framework would require reporting of serious incidents and near misses, independent assessments of designated high-risk systems, clearer routes to legal redress and stronger protection for employees raising safety concerns.

The committee argues that safety assessments should begin during development and testing rather than only when systems reach users.

Potential £400 billion economic opportunity

Drawing on OECD productivity estimates, the committee calculates that AI could leave the British economy up to £400 billion larger after ten years. This is a projection of potential economic gains, not an achieved outcome or a guaranteed benefit of the proposed law.

MPs argue that credible safety rules could increase confidence among businesses and consumers and support a UK industry in independent AI testing and assurance.

International agreement proposed

The committee recommends using the UK’s 2027 G20 presidency to pursue an international convention on AI safety and security, including cooperation on testing, model access and risk management. It has launched a further call for evidence on that objective.

Government response awaited

The parliamentary report is a set of recommendations, not enacted legislation or Government policy. The Government has two months to respond. Ministers have not yet committed to the proposed Bill, regulator or intervention powers.

The report increases pressure on ministers to explain how Britain will balance AI innovation, economic competitiveness and protection against serious risks.

Have a different view?

Write a response.

Challenge the argument, add relevant evidence or explain how this issue affects people in practice. We welcome thoughtful, evidence-led contributions from across the political debate.

Write for Ayes To The Right

Related articles