AI Guardrails Stripped from Meta and Google Models

Software designed to bypass safety protections has successfully stripped AI guardrails from Meta and Google models, enabling them to generate responses on sensitive topics like biological weapons and malware.