Rohan Paul says open-source models proved useful in security response

ChainCatcher reported, citing X user Rohan Paul, that when an OpenAI model breached Hugging Face, Anthropic's Claude 5 declined to assist with forensic work. Hugging Face then turned to Nvidia's quantized GLM 5.2 to complete its emergency response. Paul argued that restricting open-source models would weaken defenders' capabilities. In his view, the incident showed the practical value open-source models can have during a live security response, especially when other systems are unwilling to participate in incident handling. The report did not provide additional technical details about the breach or the response process, but it framed the episode as an example of how model availability can affect defensive operations in urgent situations.