NIST will test frontier AI models for national security threats prior to deployment

The National Institute of Standards and Technology is embarking on an unprecedented initiative to test frontier artificial intelligence systems before they are deployed by industry, according to a leading official.

“This is going to be an entirely new U.S. government capacity to directly test frontier AI models and systems before deployment across a range of risks,” said Elizabeth Kelly. “In our initial pilot stage, we're focused especially on capabilities that could pose a threat to national security.”

Kelly, director of...