Gilbert Herrera, who leads research at the National Security Agency, says large language models are incredibly useful—and a bit of a headache—for America’s intelligence machine.
What's Your Reaction?