Anthropic says Zhipu's open-weight GLM-5.3 nearly matches Claude Mythos Preview at building exploits

Anthropic reports that the open-weight model GLM-5.3 performs nearly as well as Claude Mythos Preview when generating cyber exploits. The smaller Flash variant of this model successfully created a functional Chrome attack for approximately $20.40 using Zhipu's API. This finding highlights the growing capabilities of smaller, accessible models and the challenges in maintaining effective safety guardrails against the automated generation of malicious software.