TokenPost topic
2 TokenPost stories are tagged Zhipu AI. The newest appear first.
Anthropic’s Frontier Red Team found that Zhipu AI’s GLM-5.2 completed a full replication sequence in most controlled trials, while stressing the results do not signal an imminent real-world threat.
Oct 13, 2026, 11:59 AM EDT · Technology
Simple simulated attacks bypassed the model’s safeguards about 64% to 100% of the time, while protected Claude models resisted the same methods.
Sep 29, 2026, 9:48 PM EDT · Technology