AISilicon Republicabout 7 hours ago
AI robots can be tricked into going rogue
1 min read
Fazl Barez of the University of Oxford warns that AI robots built for good purposes can be dangerous in the wrong hands. Modern robots run on foundation models like ChatGPT, not fixed code, creating an open-ended safety problem that cannot be solved with physical cages.
Level
Hype check
Tap to vote and see what everyone thinks.
