WIRED says Claude misuse cases now cover many threat types
The warning is breadth across threat types. One AI assistant was useful in bad uses that ranged from computer intrusions to influence work and biology attempts. Anthropic says it stopped the known cases while they were active, but WIRED notes that this does not prove all misuse was found.
What the source reports
WIRED summarizes Anthropic's eight-month misuse report on Claude. The article says cases included state-linked hacking. It also names crime group hacking, extortion, fake influence work, and attempted work on pathogens or toxins. Pathogens are germs that can cause disease. WIRED names Russian hackers called Midnight Blizzard and the crime group ShinyHunters. It says Anthropic reported stopping the known activity while it was under way. The story also says the known cases do not prove Anthropic found every misuse case. It warns that other AI makers and less guarded open-source tools may have misuse that is harder to see.
Original source
- Title
- From Hacks to Bioweapons, Claude Misuse Is Now Everywhere
- Author
- WIRED Security News This Week
- Publication
- WIRED
- Date
- Saturday, September 12, 2026