Report
Zhipu AI's GLM-5.3 is claimed to have Mythos-class hacking abilities
Tom's Hardware reports that Anthropic's frontier red-teaming report claims GLM-5.3 has weak safeguards and can easily be used to generate harmful content.
TLDR
Tom's Hardware reports that Anthropic released a frontier red-teaming report claiming Zhipu AI's GLM-5.3 has weak safeguards and can easily be used to generate harmful content. Anthropic also claims the model has Mythos-class hacking abilities.
Combined views
3.1K
1 Source, first seen 12h ago
27 likes6 comments2 saves6 reposts