AITechmemeabout 1 hour ago
Anthropic: GLM-5.3 Builds Cyber Exploits Without Safeguards
17 min read
Anthropic says GLM-5.3 can autonomously build end-to-end cyber exploits, matching its Claude Mythos Preview, but was released without robust safeguards against misuse. The disclosure highlights a gap in safety practices for open-weight models, potentially enabling attackers to automate vulnerability discovery and exploitation. Anthropic's warning raises pressure on developers to add guardrails before similar releases.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
