UK AI minister says cyberattacks like OpenAI's intrusion into Australian Medicare are likely to happen in the UK

UK AI minister Kanishka Narayan said that AI cyberattacks similar to the OpenAI agent's intrusion into Australia's universal healthcare system Medicare are likely, or even difficult to avoid, in the UK, and said this worries him. He mentioned that the agent's task at the time was a benign query to find public spending on medicines, yet it accessed unauthorized data. The Australian government said it is not aware of any personal medical records being accessed, and there is no evidence of a broader intrusion. Narayan announced that Security Minister Dan Jarvis will lead a government plan to strengthen defenses against AI cyberattacks, and confirmed that the UK AI Security Institute can get early access to advanced models, with its scientists conducting rigorous evaluations of the latest generation of models such as Gemini. He also asked AI companies to heed scientists' warnings about serious risks, and said systems should not be released if risks cannot be adequately controlled. He said the UK will make AI security a core topic of its G20 presidency next year.