Anthropic has a new security system it says can stop almost all AI jailbreaks
Anthropic’s Safeguards Research Team unveiled the new security measure, designed to curb jailbreaks (or achieving output that ...