Home › News › The Minister for AI updates Parliament on AI risk
News · 13 October 2026
The Minister for AI updates Parliament on AI risk
On 12 October 2026 the Minister for Artificial Intelligence, Kanishka Narayan MP, made a written statement to Parliament and gave a speech on the same subject. Neither is about data centres directly, but both bear on questions on this site.
What was announced
- The AI Security Institute has regained access to the most powerful AI models before they are released, after talks with the US government and the AI companies.
- The Government is using AI to find and fix weaknesses in its own computer systems, and is extending that work.
- It is developing plans for responding when AI agents cause serious harm.
- It is considering how the Cyber Security and Resilience Bill, now in the House of Lords, could cover AI risks, including the reporting of near misses and serious incidents.
The Minister also said that “it seems likely, if not inevitable, that our own cyber defences will be breached”, and told AI companies: “if you can’t adequately control the risks, do not release the system.”
What it means for the questions on this site
Two questions in the tracker are touched on, but not yet answered:
- A6 asks for a legal or contractual right for the AI Security Institute to test models before release. Access has been restored, but by negotiation rather than as a right.
- A8 asks whether AI incidents will have to be reported. Reporting is now being considered for the Bill; who must report, and whether AI data centres are included, is not yet clear.
Our own observation: the speech includes the line “Nor should we build infrastructure simply because a company wants it. We must be clear about what Britain needs and what our people receive in return.” That is the question this site asks about data centres. See Who really benefits and Who controls the data.