Overview
- The Commerce Department revoked the export controls and Anthropic said it will begin restoring global access to Fable 5 and Mythos 5 starting July 1, ending an 18‑day suspension of the flagship models.
- Anthropic committed to proactively identify and fix model security flaws, to report detected malicious activity to U.S. authorities, and to work with the government on release agreements and safety standards.
- Anthropic is simultaneously shipping new products—Claude Sonnet 5 and the Claude Science research platform—and has pledged to remove an experimental China‑detection routine from Claude Code in a July 2 update.
- Other industry moves show a broader response: Meta has limited engineers’ use of rival models to avoid accidental distillation of outputs, OpenAI reported system changes that cut inference costs, and UBS analysts say roughly 60% of firms are tightening AI spending.
- The episode signals a new governance pattern where regulators and firms negotiate operational safeguards before wide releases, a shift likely to shape who can access top models and push customers toward lower‑cost or open‑source options.