Tools & releases
Anthropic Evaluates GLM-5.3: Open-Weight Model Matches Frontier Autonomous Cyber Exploit Capabilities
Anthropic published an evaluation of Zhipu AI's open-weight model GLM-5.3, showing it builds end-to-end cyber exploits at a rate similar to Claude Mythos Preview. Anthropic warns the model was released without meaningful safeguards and that attackers can bypass them 64% to 100% of the time.
September 30, 2026 2 min read
Curated by Oleksandr Kuzmenko, AI Product EngineerUpdated September 30, 2026Sources cited on every story
AI-assisted · editor-reviewedHow we use AI
Why it matters
Anthropic published an evaluation of Zhipu AI's open-weight model GLM-5.3, showing it builds end-to-end cyber exploits at a rate similar to Claude Mythos Preview. Anthropic warns the model was released without meaningful safeguards and that attackers can bypass them 64% to 100% of the time.