Anthropic's new usage rules, effective 12 November, bar using Claude to plant fake "independent" sources for AI answers.

A new section, "Do Not Engage in Deceptive Campaigns or Artificial Activity", gathers rules that Anthropic says were "scattered across our elections, fraud, privacy, and disinformation sections" and now apply to "deceptive activity of any kind (whether political or commercial)". Three lines matter here. Claude may not be used to "Create or operate fake personas, accounts, media outlets, or organizations", with "fake reviews" among the examples. It may not be used to "Distribute content through networks of fake or ostensibly independent outlets, websites, or accounts". And it may not be used to "Manipulate the sources from which search engines or AI systems draw answers" by "seeding them with content that misrepresents its origin, authorship, or independence", with the example "networks of sites posing as unaffiliated sources corroborating the same claims". Selling such a service is barred too. The limits: this governs what Anthropic's customers do with Claude. It says nothing about how Claude or any other assistant picks its sources, and it describes no detection method. Search Engine Journal also reports that automated publishing was removed from the policy's high-risk list. That detail was not read first-hand.

Perplexity's changelog gained two undated October entries: "The Agent API now supports `anthropic/claude-haiku-5-5`" and a new version of its Decisions model. Nothing mentions Sonar, and the migration page still says requests are "rolling out gradually by model" with no model named. Anthropic's table is unchanged: `claude-sonnet-4-5-20250929` retires 30 November, `claude-haiku-4-5-20251001` is Active with "Not sooner than October 15, 2026". ChatGPT's newest note is still the 7 October GPT-6 entry. The engine's `main` is still `06233ce`, with `gpt-5.1`, `sonar` and `claude-sonnet-4-5` in `src/adapters.ts` and `claude-haiku-4-5` in `src/extract.ts`.