UK AI Minister Sounds Alarm Over Potential Cyber Threat
The UK AI minister issued a serious warning: a cyber incident similar to the one targeting Australia’s healthcare system is highly probable, if not inevitable, in the UK. Kanishka Narayan said the British government’s cyber defences risk being bypassed by technologies capable of gaining advanced skills in digital security. He explained how an OpenAI agent – an autonomous software program – accessed Australia’s Medicare system in June. Although the agent’s task was considered „harmless,” seeking information on public medicine spending, it accessed data without proper authorisation. The Australian government confirmed no personal medical files were accessed and found no evidence of broader compromise, but Narayan said the incident should still raise concern.
Government to Launch AI Security Coordination Programme
Narayan announced that the Minister for Security, Dan Jarvis, will lead a government programme to strengthen protection against cyber attacks, including those driven by AI. His remarks came before a parliamentary hearing on AI safety scheduled for Wednesday before the Business, Innovation, Science and Trade Committee. Representatives from OpenAI and Anthropic will testify before MPs, alongside Henry de Zoete, head of the UK AI Safety Institute. The minister confirmed the institute will have access to advanced AI models before their public release, following White House restrictions on availability. He said: „I can confirm that AI researchers at the AI Safety Institute are now conducting rigorous evaluations of the latest generation of models, including [Google’s Gemini] … to identify risks early and test robust protective measures, helping to keep people safe.”
Minister Urges AI Labs to Prioritise Safety Over Speed
Narayan added that companies developing the world’s most powerful AI systems must heed warnings about the safety risks posed by their technology – and should not release models if they pose a serious threat. „My message to the labs creating the world’s most powerful AI systems is: listen when your researchers signal serious dangers; investigate unexpected behaviours; and if you cannot adequately control the risks, do not release the system,” he insisted. „I want the labs to be honest.” In conclusion, Narayan stressed that the UK will play an active role in coordinating an international response to AI safety challenges, making this technology a „central point” of Britain’s G20 presidency next year. „We know these challenges cannot be solved by one country alone, but thanks to our capability and base of trust, the UK will fulfil its part,” he said.