@ArizePhoenix: it now leads CyberGym at 84.5%, and https://Z.ai is running a public disclosure ledger for the 2,400+ real-world vulner…
Summary
An AI model has achieved 84.5% on CyberGym, leading the benchmark, and Z.ai is running a public disclosure ledger for over 2,400 real-world vulnerabilities surfaced by the model.
View Cached Full Text
Cached at: 09/03/26, 06:12 PM
it now leads CyberGym at 84.5%, and https://t.co/QVncYTK2WO is running a public disclosure ledger for the 2,400+ real-world vulnerabilities the model has surfaced.
Similar Articles
@dwizzzleMSFT: http://cybergym.io just updated its leaderboard and MDASH is now #1 using a new multi-model approach. Huge credit to Ta…
Microsoft's new multi-model agentic security system (MDASH) tops the CyberGym leaderboard for vulnerability discovery, achieving 35 zero-day findings and demonstrating advanced AI-driven defense capabilities.
China’s Z.ai claims it can match Mythos on cybersecurity
Zhipu AI released its open-weight GLM-5.2 model, which reportedly matches Anthropic's Mythos in cybersecurity bug-finding tasks, narrowing the gap between Chinese and US AI capabilities.
CAISI’s Assessment of Z.ai’s GLM-5.3 Cyber Capabilities
CAISI's assessment finds that Z.ai's GLM-5.3 is the most cyber-capable open-weight AI model to date, but it still trails U.S. frontier models by approximately four months in capability.
Anthropic’s new model apparently found over 10,000 security bugs in a month
Anthropic's new AI model, Claude Mythos, identified over 10,000 high and critical security flaws in global system software within a month, with a false positive rate better than human testers, significantly advancing AI-driven cybersecurity.
@EpochAIResearch: Serious cyber vulnerability disclosures keep climbing. In July, 21 major tech organizations published ~2,500 high- and …
The tweet reports that serious cyber vulnerability disclosures are climbing sharply, with 21 major tech organizations publishing about 2,500 high- and critical-severity CVEs in July — roughly 5× the previous monthly record — following Anthropic's reveal that Claude Mythos Preview could autonomously find software vulnerabilities.