SECURITY % min read

Chinese Open-Source Model GLM-5.2 Challenges Claude Mythos in Cybersecurity

Chinese Open-Source Model GLM-5.2 Challenges Claude Mythos in Cybersecurity
Photo by Shamin Haky / Unsplash

Security researchers report that Chinese AI lab Z.ai has developed GLM-5.2, a model that can match Anthropic’s Mythos in detecting software vulnerabilities. This capability is significant because rapid bug discovery is becoming essential as companies try to stay ahead of cyberattacks. GLM-5.2 is open source, allowing anyone to download and run it locally, which makes it appealing for enterprises but also raises concerns about misuse by cybercriminals. While GLM-5.2 still trails Anthropic and OpenAI in general reasoning, its performance in cybersecurity benchmarks has narrowed the gap dramatically. The report highlights that China’s open-weight strategy contrasts sharply with recent U.S. restrictions on frontier models. Overall, the findings suggest the AI race in cybersecurity is tightening faster than expected.

Read the full story on Digital Trends →