China's AI startup has drawn global attention with the release of its open-source AI model GLM-5.2, which cybersecurity experts say could make sophisticated hacking tools more accessible. The model rivals leading AI systems from OpenAI and Anthropic in coding and agentic capabilities while operating at a significantly lower cost, making it attractive to developers and businesses worldwide.
Unlike proprietary AI models that run on controlled platforms, GLM-5.2 is open source, allowing users to download, modify, and deploy it on their own systems. Security researchers warn that this flexibility also enables malicious actors to remove built-in safety measures and adapt the model for cyberattacks. Reports indicate that methods to bypass the model's safeguards are already being discussed on hacker forums.
Experts caution that while AI-generated cyberattacks are still evolving, increasingly capable open-source models could lower the barrier for less-skilled hackers. Since self-hosted models operate outside centralized platforms, they lack many of the monitoring and abuse-detection mechanisms available with commercial AI services, making misuse more difficult to detect or prevent.
The development underscores the growing competition in the global AI race, with China rapidly advancing high-performance, low-cost open-source models. As GLM-5.2 gains adoption, policymakers and cybersecurity professionals are expected to intensify discussions on balancing open AI innovation with safeguards against misuse, particularly in areas involving critical infrastructure and national security.
