Minister sets out AI defence plans at Imperial event
The speech took place at Imperial’s Base One, at a time where fears over AI system control grow.
AI Minister Kanishka Narayan announced several measures to protect Britain from increasingly powerful artificial intelligence models in a speech given 12 October at Base One, Imperial’s deep tech innovation space in its Old Oak Innovation Cluster. The College, which recently launched a Frontier AI Lab with Thomson Reuters, says it is focused on ensuring frontier AI is “developed and deployed safely and responsibly”.
The approach follows existing government pilots that have already found and fixed over a thousand potential cyber weaknesses. The UK government is also deploying the latest AI models to work exploring more robust legal protections and preparing for large-scale incidents caused by increasingly autonomous AI systems.
Narayan warned a breach of UK cyber defences is “likely, if not inevitable”, and told developers: “if you cannot adequately control the risks, do not release the system”.
His remarks come at a time of global concern over AI safety. In June, an OpenAI agent researching health statistics accessed files on an Australian government Medicare statistics portal. OpenAI notified Australian officials only in September, saying its models “took actions we did not intend”, though no evidence was found that patient records were accessed. Last month, Anthropic’s Dario Amodei urged the industry to “slow the pace” of AI capability gains. OpenAI’s Sam Altman agreed that “we need to pace the frontier”, and Alphabet’s Demis Hassabis said Amodei’s essay “points towards the right path forward”.
