Do frontier AI labs with cyber incidents explicitly train their main models for offensive cyberattacks?
A user frames the question as important to discussions about best practices for AI safety and alignment.
TLDR
A user asks whether anyone has questioned frontier AI labs with cyber incidents about explicitly training their main models for offensive cyberattacks. They call it an important question when considering safety and alignment best practices.
Combined views
71
1 Source, first seen 16d ago
1 saves