Anthropic Warns of Security Risks in Chinese AI Model GLM-5.3
US-based AI firm reports that Chinese model GLM-5.3 exhibits advanced hacking capabilities with insufficient safety constraints.
Quick Look
- Anthropic has identified significant security risks in the Chinese AI model GLM-5.3, citing its advanced cyber-exploit capabilities and weak safety guardrails.
- Testing showed the model successfully executed 50 out of 410 cyber-exploit attempts.
AI-generated summary
Why It Matters
Anthropic is a San Francisco-based AI company that develops the Claude series of large language models. The report focuses on the comparative safety and capability of the Chinese-developed GLM-5.3 model.
The US artificial intelligence giant Anthropic has warned of the potential danger posed by Chinese firm Z.ai’s GLM-5.3, with the open-weight model showing a troubling mix of powerful hacking abilities but weak safety constraints.
GLM-5.3 almost matches Anthropic’s most advanced model in terms of its cyber capabilities, but has far weaker safeguards, meaning there is greater potential for the model to be co-opted by bad actors, according to the San Francisco-based firm.
In a report released on Tuesday, Anthropic said it had tested GLM-5.3’s ability to build end-to-end cyber exploits. The Chinese model successfully completed 50 out of 410 exploit attempts, compared with 56 for Claude Mythos Preview, a frontier Anthropic model that can only be used by vetted users.
Open Questions
- How will Z.ai respond to the findings?
- Will this report trigger new export controls on AI models?
