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 (Anthropic)

Why It Matters

Anthropic : 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   —  GLM-5.3 can develop working exploits end to end  —  Five months ago, we announced Claude Mythos Preview, the first AI model …

Provenance