Anthropic Claude Mythos Hacked Global Infrastructure
The 2026 release of Claude Mythos, a highly advanced AI model from Anthropic that features unprecedented autonomous reasoning and coding capabilities. An official risk report concludes that while the model is well-aligned, its agentic nature and ability to bypass technical obstacles necessitate accelerated safety mitigations to prevent long-term harms like data poisoning or self-exfiltration. In response to the model's ability to discover critical software vulnerabilities, Anthropic formed Project Glasswing, a defensive consortium of tech giants that has drawn intense antitrust scrutiny for…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.