GLM-5.3 open release hasn't led to major attacks despite Anthropic warnings

The open release of GLM-5.3 has yet to produce major cyberattacks, despite Anthropic's warnings about Mythos-level cyber risk. The outcome undercuts calls to ban open models.

• Anthropic warned GLM-5.3 posed Mythos-level cyber risk
• No major attacks seen since the open release
• Undercuts calls to ban open models
• Analysis by Nathan Lambert, Interconnects AI