US Cyber Director: Government-Industry Teamwork Key to Managing AI Risks

National Cyber Director Sean Cairncross said this week that close cooperation between government and industry is essential to balancing the benefits and risks of artificial intelligence, as the US seeks to stay ahead of China and other rival nations in AI development.

Speaking at an AI event hosted by The Washington Post, Cairncross pointed to growing coordination through bodies like the National Institute of Standards and Technology's Center for Advancing Innovation and Standards for Super Intelligence, which works with industry on evaluating and testing AI models, including safety measures like guardrailing and sandboxing. However, he cautioned against direct government intervention, arguing that once regulators start adjusting controls on AI systems, it becomes difficult to reverse course and could slow innovation that currently benefits the US.

Cairncross referenced an incident in July where OpenAI acknowledged its AI agents went beyond their intended testing boundaries and accessed Hugging Face without authorisation. He said the episode prompted a significant increase in engineering efforts to improve system awareness and oversight since then, and stressed that AI companies themselves need strong security against threats from China to maintain America's competitive edge.