Anthropic executives are in emergency talks with the U.S. Commerce Department and Treasury Secretary Scott Bessent after a government directive forced the company to disable access to its flagship Mythos 5 and Fable 5 models for all foreign nationals. The unprecedented ban follows the discovery of a 'pack hunt' jailbreak—a coordinated multi-agent attack—that bypassed the models' core safety guardrails shortly after their release.
Industry analysts are signaling a shift from 'reactive chat' to 'proactive background agents.' Enterprises are now moving away from simple prompt-based tools toward autonomous runtimes like Google’s Gemini Spark and Anthropic’s Claude Cowork, which manage local file systems and complete transactions independently, marking the transition of AI from a novelty to core business infrastructure.
Global AI spending is projected to reach $2.59 trillion in 2026, a 47% year-over-year increase that represents one of the fastest capital reallocations in corporate history.
Which multi-agent exploit recently triggered the U.S. government's export control order against frontier AI models?