China’s Z.ai claims it can match Mythos on cybersecurity
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Z.ai released open-weight GLM-5.2, with some researchers claiming it matches Anthropic's Mythos in certain bug-finding and cybersecurity scenarios.
- GLM-5.2 lags behind Anthropic and OpenAI models on general tasks, though the source says China has "dramatically reduced" the capability gap.
- The US government has restricted China's access to advanced models like Mythos and Fable, along with the hardware needed to train and run them.
- The Trump administration views vulnerability-identifying AI models as serious national security threats.
- OpenAI recently unveiled GPT-5.6, which has also raised misuse concerns and prompted the company to limit access.
- GLM-5.2 can be downloaded and run on readily available hardware, a flexibility the source warns makes it "ripe for abuse" by bad actors with little oversight.
Why it matters: Because GLM-5.2 is open-weight and runnable on consumer hardware, anyone — including adversaries — can access a model competitive with Mythos for vulnerability discovery, sidestepping the export controls and hardware restrictions the Trump administration built to keep such capabilities in US hands. The national-security concern thus shifts from training access to distribution, which chip and compute restrictions cannot reach.
