Jen Easterly: Guard Must Be Built Early In Artificial Intelligence Development

Jen Easterly, director of the United States Cybersecurity and Infrastructure Agency. (photo: x @RIC_CyberInst)

A top US official said Monday 27 November that the potential threat posed by the rapid development of artificial intelligence (AI) means protection should be built into the system from scratch rather than later added.

"We already consider it natural a world where technology products come out of production lines full of vulnerabilities, and then consumers are expected to improve those vulnerabilities. We cannot live in such a world with AI," Jen Easterly, director of the United States Cybersecurity and Infrastructure Agency.

"AI is too strong, moving too fast," he said in a telephone interview after holding talks at Ottawa with Sami Khoury, head of Canada's Cybersecurity Center.

Easterly spoke on the same day institutions from 18 countries, including the United States, supporting the UK's newly developed AI cybersecurity guidelines, which focus on design, development, implementation, and safe maintenance.

"We must consider security throughout the life cycle of AI capabilities," said Khoury.

Earlier this month, leading artificial intelligence developers agreed to cooperate with the government in testing new frontier models before they were released, in order to help manage the risks of this fast-growing technology.

"This month, leading AI developers agreed to cooperate with the government to test a new frontier model before its release to help manage the risks of this fast-growing technology," said Easterly.