Constitutional Classifiers: Defending against Universal Jailbreaks...#Anthropic#Classification#Safety#Large Language Models#Paper#PDF·arxiv.org·Feb 3, 2025Constitutional Classifiers: Defending against Universal Jailbreaks...
Simple probes can catch sleeper agents \ Anthropic#Training#Large Language Models#Anthropic#Paper#Classification#Cybersecurity·anthropic.com·Apr 24, 2024Simple probes can catch sleeper agents \ Anthropic