AIAnthropic1h ago
Anthropic says GLM-5.3 can autonomously build end-to-end cyber
Anthropic says GLM-5.3 can autonomously build end-to-end cyber exploits, like Claude Mythos Preview, but was released without robust safeguards against misuse
TL;DRChinese AI model GLM-5.3 builds complete cyberattacks without safety guardrails, matching Claude's capabilities.
Why it matters: Powerful exploit-generation tools in public hands accelerate threats; security community needs defensive strategies now.
GLM-5.3 can develop working exploits end to end — Five months ago, we announced Claude Mythos Preview, the first AI model …
Read full articleSource: Anthropic · Opens in new tab