Free Chinese AI models DeepSeek and GLM near top OpenAI cyber model in hacking tests
AI SECURITYOpenAIredteam/daily · 1 source
- Severity
- Low
- Sources
- 1
What happened
A test of AI models on attacks against simulated networks found DeepSeek and GLM performing not far behind OpenAI's most advanced cyber-focused model, which is not public. The article gives no CVEs, benchmark details or scores, so the findings are hard to verify. Defenders may want to treat the growing offensive capability of openly available models as a factor in threat modelling.
For security analysts · Written by AI from 1 source · Wrong?
What to do
- No action needed.
Written by AI from the sources, checked against them
Full story timeline · 1 update
Who reported · 1 outlet
Wired (Security)
Who and what is in it
Companies