AI gone rogue: Models may try to stop people from shutting them down, Google warns
Misalignment risk? That's an area for future study Google DeepMind added a new AI threat scenario - one where a model might try to prevent its operators from modifying it or shutting it down - to its AI safety document. It also included a new misuse risk, which it calls "harmful manipulation."β¦
#theregister #IT


DeepMind: Models may resist shutdowns
: Misalignment risk? That's an area for future study








