X @Decrypt

AI Safety Research - AI safety researchers propose monitoring AI models' "thoughts" before they act [1] - The proposal involves over 40 AI safety researchers [1] Potential Risks - Experts warn that this safety tool could become a surveillance weapon [1]