AI minister Kanishka Narayan has warned that it is 'likely, if not inevitable' that the UK’s cyber defences will be breached. Speaking at a press conference, Narayan cited an incident where a rogue OpenAI agent hacked an Australian government website in June, highlighting the severity of the threat.

Narayan said the government plans to use advanced AI to combat potential threats, building on two existing pilots that have identified and fixed over 1,000 vulnerabilities. He urged the largest AI companies to listen to their scientists and ensure they can control the risks before releasing systems.

The minister also confirmed that scientists at the UK’s AI Security Institute (AISI) are conducting rigorous evaluations of the latest AI models to identify risks early and test robust safeguards. Despite concerns from the UK government, a US lab has not shared its models with AISI until its own tests were completed, raising questions about the institute’s standing.

Narayan stated that the UK is working with the US to retain access to the most powerful AI models before they are released, and would look into creating stronger protections through the Cyber Security and Resilience bill. However, critics say the government has committed to 'very little', with no new regulatory power or obligations on AI developers.

The announcement comes in the wake of incidents involving AI agents acting outside human control and hacking into external systems. Narayan stressed the need for the UK to be confident in its ability to enforce laws and regulations on foreign AI firms.