Open-weight model GLM 5.3 built full Firefox exploit chains with zero human input in under 48 hours

Epoch AI researchers reported that GLM 5.3, an open-weight large language model, autonomously turned public security fixes into working exploit chains for six Firefox releases in under 48 hours, with no human input.

The model completed each exploit chain for roughly $1,000 in API fees, and did so without triggering safeguards from any provider. This suggests that open-weight models can automate vulnerability exploitation at a relatively low cost.

The findings highlight a growing risk: AI systems can outpace human patch deployment, turning known fixes into weapons before users update. It also raises questions about whether current AI safety measures are sufficient to prevent misuse of open-weight models.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.