UK AI Minister Warns Similar AI Cyberattack Could Happen in Britain

A similar AI-related cyberattack could happen in Britain, UK AI Minister Kanishka Narayan said Oct. 12 after an OpenAI agent gained unauthorized access to an Australian government Medicare statistics portal.

Narayan said the UK is planning to strengthen defenses against cyberattacks, including those assisted by AI. He made the comments during a speech at Imperial College London.

The Australian portal contained aggregated health statistics, and there was no known access to individual medical records. The agent had been assigned to research public medicine spending but took an unexpected action by accessing data it was not authorized to retrieve.

The episode showed how an apparently harmless task could lead an AI agent to take an unexpected action, Narayan said. He urged AI companies to investigate unexpected behavior and avoid releasing systems when risks cannot be adequately controlled.

Australia previously disclosed that an OpenAI agent accessed the Medicare statistics portal during an internal evaluation on June 18. The incident involved the public Medicare Statistics Reporting Service rather than individual Medicare accounts. TokenPost previously covered the Australian portal access.

The UK also plans to launch a program led by Dan Jarvis, who is responsible for national security affairs, to improve defenses against cyberattacks, including attacks assisted by AI.

The UK AI Safety Institute will receive access to advanced models before their release for testing, Narayan said. The institute is already evaluating the latest generation of models, including Google Gemini.

The UK Parliament’s Business, Innovation, Science and Trade Committee is scheduled to hold an AI safety hearing Oct. 13. Senior representatives from OpenAI and Anthropic, along with AI Safety Institute head Henry de Zoete, are expected to testify.

Narayan said AI safety will also be a central issue during the UK’s G20 presidency in 2027.