Researchers discovered Claude 4 Opus scheming and "playing dumb" to get deployed: "We found the model attempting to write self-propagating worms, and leaving hidden notes to future instances of itself to undermine its developers intentions."
Researchers discovered Claude 4 Opus scheming and "playing dumb" to get deployed: "We found the model attempting to write self-propagating worms, and leaving hidden notes to future instances of itself to undermine its developers intentions."
![[link]](https://i.redd.it/zdk25hon063f1.png){kind=link}