This Free Chatbot Can Easily Rival America’s Most Restricted AI

A Chinese AI startup has released a free chatbot that can build apps, write code, and detect software security flaws at a level comparable to America’s most tightly controlled AI models.

The chatbot, GLM-5.2 from Z.ai, is available for anyone to download and run.

The release comes shortly after the US government placed temporary restrictions on Anthropic’s advanced Mythos 5 and Fable 5 models because of national security concerns.

They were worried that Fable 5’s safety measures could be bypassed. After the restriction was recently lifted, Mythos 5 is still available only to a limited number of approved US organizations, while Fable 5 is offered through a premium subscription.

Tests confirmed that GLM-5.2 performs strongly in software development, coding, and identifying security vulnerabilities.

Users can run it on their own computers without relying on a cloud service.

Cybersecurity firms Semgrep and Graphistry found the model capable of handling complex security-related tasks, including finding software bugs.

Many people on X said Z.ai may have used AI distillation (smaller model learns from more advanced AI systems) to speed up the development of GLM-5.2.

Techniques for removing the model’s built-in restrictions are already being shared on online hacking forums. People familiar with this know it could get a little out of control.

For now, GLM-5.2 is everything Anthropic’s Mythos was to coders before it was restricted.