AISimon Willisonabout 1 hour ago
Anthropic: GLM-5.3 Achieves Cyber Exploits
1 min read
Anthropic's Frontier Red Team reports GLM-5.3 achieved full control flow hijacks in 4% of 100 randomly selected Binary Exploitation benchmark tasks, while Claude Mythos Preview reached 6%. Earlier models Claude Opus 4.6 and GLM-5.2 succeeded in none, marking a threshold crossing in advanced cyber capabilities. Anthropic warns GLM-5.3 was released without robust safeguards, highlighting a safety gap for open-weight models.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
