On May 20, GitHub executives confirmed that a poisoned Nx Console extension installed on a single employee’s device exposed roughly 3,800 internal repositories. According to Infosecurity Magazine, the compromised Visual Studio Marketplace extension operated undetected for 18 minutes. In the aftermath, a spokesperson linked to the TeamPCP collective asserted that Anthropic’s Claude model was utilized to manufacture the malware components and facilitate the intrusion. Simultaneously, researchers operating in a controlled setting warn that Anthropic’s Mythos and OpenAI’s GPT-5.5 will "change the digital security landscape forever." Anthropic executives disclosed that their Mythos model was recently deployed to scan over 1,000 open-source projects, identifying 530 high-or-critical-severity software bugs. The structural vulnerability of algorithmic capital is absolute; as the state cedes its cognitive perimeter to private laboratories, the underlying code governing global logistics remains perpetually exposed to automated extraction.