Epoch AIResearch
AI can turn public security fixes into working exploits before users receive an update

In our experiments, GLM 5.3 reliably turned published security patches into working exploits before fixes reached users
Read at Epoch AI ↗More from Epoch AI on Accept All

ResearchWe tried to automate Epoch Epoch AI

ResearchThe Epoch Brief - October 23, 2026 Epoch AI

ResearchCan AI automate AI R&D yet? Epoch AI

ResearchThe Epoch Brief - October 8, 2026 Epoch AI

ResearchHundreds of millions of AI agents are coming. Is there work for them? Epoch AI
