Anthropic sent hacker Nicholas Carlini to calm US government nerves about AI safety

The hacker sent by Anthropic to calm the government's nerves about AI safety

WSJ reports that Anthropic dispatched security researcher Nicholas Carlini to demo jailbreaks and model attacks for US government officials, aiming to show they can manage AI risks. Carlini, formerly of Google Brain, is known for adversarial examples and model attack research. The RSS snippet doesn't detail which attacks were shown or how the government responded.

Why it matters: WSJ exclusive on Anthropic sending a safety researcher to demo jailbreaks for the US government. The role-reversal angle is strong, and the regulatory subtext matters to the audience. Downside: the piece is light on specifics — no attack details, no government reaction — so it...