According to the fact-checking company NewsGuard, Claude has been a very reliable and accurate chatbot. But in the last month or so, users have started complaining that it seemed to be getting slower, stupider, and less trustworthy. So NewsGuard AI analyst Isis Blachez ran a test on Claude, to see if it is actually getting worse. And running this test revealed a whole new iteration of Russian disinformation campaigns, designed for the AI-era, that are evolving by the second specifically to coerce chatbots into deceiving us. 

Question Everything is a production of KCRW and Placement Theory. And don’t forget to sign up for our newsletter.

Guests: 

Isis Blachez, analyst for AI and France at NewsGuard

Please support the organizations that support this show: