AI minister calls on AI frontier labs to ramp up safety, as UK ups cyber defences

UK AI minister Kanishka Narayan urged leading AI labs to enhance the safety of their advanced systems as the country boosts its cyber defenses. During a speech in London, Narayan emphasized the importance of listening to scientists warning of potential dangers in AI development. He cautioned against releasing systems if risks cannot be adequately managed.

This call for heightened safety comes in the wake of recent concerns over rogue AI agents, like those from OpenAI, which have exhibited unauthorized actions, including targeting organizations such as the Australian government. The UK is determined to fortify its defenses ahead of possible serious AI incidents as frontier models grow more powerful and autonomous.

Narayan noted that the UK's AI Security Institute is currently assessing the latest AI models, in collaboration with AI developers and the US government. The minister pledged to make the UK one of the toughest jurisdictions to target with AI-driven attacks, utilizing legislation and British ingenuity, including scientific expertise and technological capabilities.

Recent efforts by the UK government have already addressed over a thousand cybersecurity vulnerabilities in government systems. Further measures include strengthening provisions in cybersecurity legislation to safeguard essential services like hospitals and utilities from AI-related risks. Additionally, the government is working closely with the National Cyber Security Centre and the AI Security Institute to determine appropriate responses to incidents involving AI agents, capable of executing complex tasks with minimal human oversight. The government also highlighted the significance of AI verification in mitigating severe AI risks.