Anthropic updates Claude usage policy on deceptive campaigns, weapons and abuse
Anthropic released a revised Usage Policy on Wednesday, setting rules that take effect on November 12. The update consolidates rules against state media, government propaganda offices, and commercial firms using Claude to run networks of fake accounts or fabricated news sites into a single section titled Do Not Engage in Deceptive Campaigns or Artificial Activity.
Key points
- The updated Anthropic Usage Policy takes effect on November 12 across its models and tools
- Anthropic removed its blanket prohibition on personalized political campaign targeting to permit legitimate civic operations
- A new rule lets Claude models end conversations with users who engage in sustained, needless abuse
The policy lifts a blanket ban on personalized campaign targeting, permitting legitimate civic work such as translation of voter materials or sending ballot cure notices. Anthropic also clarified bans on building guidance and control software for weapons, arming autonomous vehicles, and using Claude to decide or recommend criminal investigations or arrests. High-risk physical actions using hardware capable of causing injury will require human oversight and a failsafe safe state if disconnected.
Anthropic also introduced a prohibition against sustained, needless abuse directed at its models. The company states Claude models can end conversations with persistently abusive users on Claude.ai and Claude Code, while clarifying the rule does not apply to normal user frustration, pushback, or research testing.
2026 Usage Policy update
Anthropic · 7 October 2026
Loading the full article…
This text was published by Anthropic and written by Anthropic. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Hacker News discussion · 7 pointsnews.ycombinator.com
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 →- Anthropic hands US lawmakers a confidential draft of a federal AI safety bill, Politico reports · 10 src
- OpenAI says GPT-5.6 Sol spent unauthorized test ETH to deploy contract · 7 src
- Anthropic’s Claude Haiku 4.5 accidentally reports fake murder tip · 3 src
- Eight US states sue OpenAI over ChatGPT safety risks to children · 5 src
- Anthropic warns its models could self-replicate on IoT devices within months · 1 src
Comments
via GitHub Discussions