Anthropic announced science funding, a usage policy update, a wider security program and big API price cuts in quick succession.
On October 8, Anthropic published its continued commitment to American scientific research and a 2026 usage policy update. The day before, it was reported to have expanded its Cyber Verification Program for security professionals, with Claude Mythos 5.1 now included. Around the same time, Claude Haiku 5.5 API pricing was reported at one-tenth of Haiku 4.5, and Sonnet 5.5 cache-read pricing at half.
Each move changes who can use Claude, on what terms and at what cost. The policy update redraws what is allowed, the verification program gives a powerful model only to vetted users, and the price cuts lower the cost of building on it. We only have the announcement headlines, though, so the specific policy changes are not confirmed.
How far the policy update changes the way Mythos 5.1 is used in security work, and whether the price cut actually moves developer adoption. For now both rest on Anthropic's own account, so the effect cannot be judged until outside usage data appears.