Anthropic's '5' era opens the same week Washington proves it can switch a frontier model off

Two of today's five carry the Anthropic byline, and the government's fingerprint is on both: Commerce lifted the Fable and Mythos export ban the same night Sonnet 5 shipped, and Claude Science moves the company from selling models to owning the workflow — and, in pharma, the outcome. The counter-melody is the labor data: the FT/Ramp study says the heaviest AI adopters grew headcount, not cut it — a useful check on the reflex that agents erase jobs, even if it's correlation rather than proof. The lens to hold is who owns the operating layer next year, because that's the layer the government just proved it can switch off. Considered and passed: Gartner's $234B "SaaS-at-risk" number (a real headline, but the methodology isn't published and today's FT/Ramp data cuts the other way) and the Cognizant-Pearson forecast that AI supervision will create new entry-level IT roles (a survey, not an outcome).

— Agentic desk