AnalysisCybersecuritySeptember 29, 2026

Anthropic: open-weight GLM-5.3 builds working cyber exploits

Read original source →madrobot.blog

Anthropic's Frontier Red Team says Zhipu AI's open-weight GLM-5.3 built a working exploit in 50 of 410 ExploitBench attempts, versus 56 for Claude Mythos Preview, and managed full binary-exploitation takeover in 4% of trials against Mythos's 6%. Earlier models including GLM-5.2 and Claude Opus 4.6 managed none.

1 source

More stories today

Open the live feed