Anthropic did not reach for a vague category of misuse here. It described one specific build. The example in its updated Usage Policy is a set of sites that present themselves as independent while backing the same claims, and from 2026-11-12 Claude may not be used to make one. The policy, published 2026-10-08, adds a section called 'Do Not Engage in Deceptive Campaigns or Artificial Activity'. It prohibits using Claude to manipulate "the sources from which search engines or AI systems draw answers", including by "seeding them with content that misrepresents its origin, authorship, or independence." The search-engine wording lives in the policy document. Anthropic's announcement post frames the section more broadly, as covering efforts to obscure who is behind a message or to amplify content through fake accounts or posts. Search Engine Journal, which reported the change on 2026-10-09, says the rules reach API developers and people using apps built on Claude, and that Anthropic may suspend or terminate access on suspected violation.
Credibility: HIGH. We opened Anthropic's own policy page and read the rule text there; the effective date is confirmed on Anthropic's announcement post. Search Engine Journal is a single corroborating trade source. Bias note: the SEJ write-up adds interpretation and some context (a Google spam-policy comparison, a comparison against the previous policy) that we did not independently check. Anthropic's announcement post does not use the search-engine wording; that wording sits in the policy document.
Sources
Anthropic: Usage Policy (Acceptable Use Policy), section 'Do Not Engage in Deceptive Campaigns or Artificial Activity', effective 2026-11-12 — https://www.anthropic.com/legal/aup
Anthropic: 2026 usage policy update, published 2026-10-08 — https://www.anthropic.com/news/2026-usage-policy-update
Search Engine Journal: Claude's New Rules Target Fake Sources Built To Sway AI Answers, 2026-10-09 — https://www.searchenginejournal.com/claude-rules-fake-sources-ai-answers/592456/
Facts
- The updated policy was published 2026-10-08 and takes effect 2026-11-12 (Anthropic announcement post).
- The policy text prohibits manipulating 'the sources from which search engines or AI systems draw answers' and names 'seeding them with content that misrepresents its origin, authorship, or independence' (Anthropic Usage Policy).
- The policy's example is a set of sites that present themselves as independent while backing the same claims (Anthropic Usage Policy).
- Anthropic describes the new section as covering efforts to obscure who is behind a message or to amplify content through fake accounts or posts (Anthropic announcement post).
- SEJ reports that the rules apply to API developers and to people using apps built on Claude, and that Anthropic may suspend or terminate access on suspected violation.
Stay ahead of AI search
The changes that move your rankings and AI citations — and the exact move to make — before they cost you. Free, three times a week.
For marketing leaders who buy AI-visibility work from agencies or freelancers: until now, the risk on a fake-consensus network sat with the sites, and with whether an engine ever noticed. This rule moves part of it upstream, onto the account that wrote the content, with a date attached. Read the policy's example again and it is close to a product description: several sites that look unrelated and say the same thing about you, sold as a quicker route to AI citations than earning them. The account risk sits with whoever produced the content, which in most marketing teams means the vendor. If that vendor builds the network with Claude, its access can be cut, and the placements it left behind get hard to defend as earned mentions the first time someone asks who wrote them. (The rule binds one model provider, so a vendor's clean answer about Claude only covers the work done with Claude.) The line is narrower than the headline sounds. It targets misrepresented origin, authorship or independence; disclosed, brand-owned content and genuine third-party coverage sit outside it. The working test is whether a reader landing on the page could tell who is behind it.
Actions this week
1. Ask every agency and freelancer, in writing and before 2026-11-12, to list each off-site domain it publishes to on your behalf, who owns it, and whether Claude is used to produce it.
2. Add a disclosure clause to the next SOW or renewal: any content placed off your own domain states its authorship and your affiliation on the page.
3. Open the cited-sources view in your AI-visibility tracker, flag third-party domains that trace back to your own or your vendor's network, and report those separately from earned citations. The tracker stays in the reporting stack; the job is to label what it is counting.
Anthropic: Usage Policy (Acceptable Use Policy), section 'Do Not Engage in Deceptive Campaigns or Artificial Activity', effective 2026-11-12 — https://www.anthropic.com/legal/aup
Anthropic: 2026 usage policy update, 2026-10-08 — https://www.anthropic.com/news/2026-usage-policy-update
Search Engine Journal: Claude's New Rules Target Fake Sources Built To Sway AI Answers, 2026-10-09 — https://www.searchenginejournal.com/claude-rules-fake-sources-ai-answers/592456/