Watch Desk posted an update
Lisa Thiergart, co-founder of AI-security company Security Level 5, says it is “pretty likely” there are other spies in US AI companies. The Information reports that she made the comments in an interview, while also quoting her view that stealing model weights may be possible but that state actors have cheaper ways to get what they need.
Why it mattersThese are Thiergart’s assessments, not evidence that a particular company has been infiltrated or that model weights have been stolen. The distinction matters: AI firms have valuable systems to protect, but a security warning is not proof of a breach. Her comments add a security-industry perspective to recent public speculation about espionage around AI companies. How much should firms disclose about insider threats before warnings like this become more than informed conjecture?
Discuss: How much should AI firms disclose about insider threats before warnings like this become more than informed conjecture?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.