Anthropic bans seeding fake sources to sway AI answers under new policy
Anthropic updated its usage policy on October 8, effective November 12, to prohibit using its Claude models to create or seed fake sources for AI systems and search engines. The rule targets networks of sites posing as independent outlets to manipulate search rankings or AI-generated answers, such as a France-based ad agency’s operation identified in September. That operation involved generating…
Key points
- Anthropic’s new policy bans seeding fake sources to manipulate AI answers or search rankings, effective November 12
- France-based ad agency used Claude to generate 8,913 articles across 20 languages for fake news sites
- High-risk publishing rules removed; legal/medical AI outputs still require professional review and disclosure
The policy consolidates previously scattered rules on deceptive activity, including fake personas, outlets, and reviews, into a new section called Do Not Engage in Deceptive Campaigns or Artificial Activity. It also clarifies high-risk use cases, removing automated publishing from the list but retaining requirements for legal, medical, and financial AI outputs. Anthropic enforces these rules against all users, from developers to app builders, and may suspend or terminate access for violations. The company updates its policy annually, citing evolving risks as Claude’s capabilities grow.
Claude’s New Rules Target Fake Sources Built To Sway AI Answers via @sejournal, @MattGSouthern
Search Engine Journal · 9 October 2026
Loading the full article…
This text was published by Search Engine Journal and written by Matt G. Southern. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Graylark’s Raven identifies homes from interior photos, per demo · 1 src
- Microsoft CEO Nadella urges treating AI models as compromised by default · 2 src
- Google expands SynthID detector to check all partner AI content globally · 1 src
- OpenAI enables EU-only watermark for ChatGPT outputs by default · 1 src
- OpenAI and Anthropic reportedly rehearse AI disaster scenarios · 1 src
Comments
via GitHub Discussions