In a Usage Policy update published October 8, 2026 and effective November 12, 2026, Anthropic added a section — "Do Not Engage in Deceptive Campaigns or Artificial Activity" — that explicitly prohibits using Claude to plant content in the sources search engines and AI systems draw answers from when that content hides its origin, authorship, or independence. The rule binds people and companies using Claude, not search engines, and follows Anthropic's own report on a network of roughly 70 fake local-news sites.
2026-10-10 · by Moe Ameen
Anthropic published an updated Usage Policy on October 8, 2026, with an effective date of November 12, 2026, and the notable addition for anyone doing content at scale is a section titled "Do Not Engage in Deceptive Campaigns or Artificial Activity." It bans using Claude 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. The policy's own example is a network of sites posing as unaffiliated sources that corroborate the same claims.
The section is broader than that one line. It also prohibits creating fake personas or media outlets to mislead audiences, distributing content through networks of fake or ostensibly independent outlets, websites, or accounts so their shared source stays hidden, and building tools for — or selling as a service — this kind of coordinated inauthentic behavior. The prior Usage Policy, last updated in 2025, did not mention search engines or AI answers at all. The rule binds the people and organizations using Claude, not the search engines themselves, and Anthropic enforces it through the usual ladder of warnings, usage limits, suspension, and account termination.
The change did not come from nowhere. Anthropic's earlier threat-intelligence reporting described a France-based operation that ran roughly 70 websites dressed up as independent local-news outlets and used Claude to mass-produce articles in a consistent template — each with three to four internal links — explicitly to boost the sites' authority signals with search engines. Anthropic says at least 8,913 articles were published across roughly 20 languages, that most got little real engagement, and that it removed the Claude account and banned the organization behind it.
A note on framing: some coverage points out that Anthropic's announcement post emphasizes the broader deceptive-campaign rules and does not foreground search engines, so if the exact wording matters for your use case, read the Usage Policy text directly. Treat the threat-report figures (site count, article count, languages) as Anthropic's own numbers rather than independently audited ones.
The practical question this raises for a legitimate creator is: if manufacturing fake independent sources to show up in AI answers is now banned, what is the sanctioned way to earn those citations? The answer is the unglamorous one — publish genuinely good, clearly-attributed content under your own identity, consistently, across the surfaces AI systems actually read. That is the job [Kompozy](/) is built for. One source becomes a spread of on-brand pieces — a [Persona Short](/glossary/persona-shorts) and avatar video, carousels, quote graphics, native text posts, a [blog article on your own domain](/ai-tools/claude-fake-source-rules), and an email newsletter — all governed by a [Persona Brief](/glossary/persona-brief) so every piece carries one transparent voice that is unmistakably yours, not a laundered persona.
The distinction the policy enforces is the one Kompozy was already designed around: credibility you own versus credibility you fake. Instead of 70 disposable sites quietly linking to each other, you build one real brand posting disclosed content to the eight social platforms plus your blog and email, with a per-post review gate before anything ships and [Autopilot](/glossary/autopilot) keeping a steady cadence behind it. That is what AI Overviews and ChatGPT-style answers increasingly cite — a consistent, attributable source — and it is the opposite of the throwaway-network tactic Anthropic just outlawed. For the surrounding story, see [OpenAI's takedown of a ChatGPT fake-byline campaign](/news/openai-iran-chatgpt-fake-articles-campaign), [Google's warning on fake AI author attribution](/news/google-warns-fake-ai-author-attribution), and [which sources AI platforms actually cite in 2026](/news/ai-platform-citation-source-index-2026).
In a policy published October 8, 2026 (effective November 12, 2026), Anthropic added a section called "Do Not Engage in Deceptive Campaigns or Artificial Activity." It prohibits using Claude to manipulate the sources search engines or AI systems draw answers from by seeding them with content that misrepresents its origin, authorship, or independence, and it bans fake personas, fake outlets, networks of ostensibly independent sites, and selling such campaigns as a service.
No. The rule targets deception, not AI writing. Publishing AI-assisted content under your own brand and name is unaffected. What is banned is engineering content to hide who created it — for example, a network of look-alike "independent" sites built to corroborate your own claims and inflate search or AI-answer authority.
It applies to the people and companies using Claude, including developers on the API and apps built on Claude — not to search engines themselves. Anthropic enforces it with its standard ladder: warnings, usage limits, suspension, and account termination if it finds a violation.
It follows the company's threat-intelligence reporting on a France-linked operation that ran roughly 70 fake local-news sites and used Claude to mass-produce templated articles — each with three to four internal links — to boost the sites' search authority. Anthropic says at least 8,913 articles ran across about 20 languages before it banned the organization.