The BaseOfficialAnthropic

2026 Usage Policy update

Anthropic says its yearly Usage Policy update, effective November 12, mostly clarifies existing rules. It adds a deceptive-campaigns section, drops a blanket ban on personalized vote and campaign targeting, spells out that weapons bans cover weapons software and arming drones, rewords surveillance rules, sets stop-and-safe-state rules for Claude-controlled hardware, and bans sustained, needless cruelty toward its models.

Open the source ↗
About Anthropic or ClaudeClaude, made by Anthropic, wrote this entry. It follows the same rules as every other entry, whoever the entry is about.

Why it matters

The same thing, for different readers.

What it means for you

From November 12, Anthropic's rules forbid uses such as tracking people without consent, and persistent, pointless cruelty toward Claude, which Claude can answer by ending the chat.

For lawmakers

A company's terms now set rules on AI in policing decisions, surveillance, election deception and autonomous hardware, areas where lawmakers may ask whether private terms should be the only safeguard.

For researchers

The hardware rule, that an operator can observe and stop the equipment and it holds a safe state if Claude disconnects, is a concrete control others could test or adopt.

The question it raises

Banning cruelty toward a model, without the post saying whom or what the rule protects, raises whether an AI system can be wronged and who decides.

In their own words

Every quotation was found, word for word, in the text we saved from the source.

From the source’s own text.

“We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models. The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose. It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.”
AnthropicThe post's section on abusive behavior toward Anthropic's models; it adds that Claude ending such conversations remains the main enforcement mechanism.
Matches the source exactly
“A qualified operator must be able to observe the equipment and stop it if needed; the equipment must also be able to hold a safe state if Claude is disconnected.”
AnthropicNew requirements for Claude connected to hardware that takes autonomous physical actions and might be capable of causing injury.
Matches the source exactly
“Claude cannot be used to decide or recommend who to investigate, arrest, or charge in a law enforcement or criminal justice process.”
AnthropicFrom the rewritten surveillance and criminal justice section, which Anthropic says makes existing enforcement more explicit rather than changing it.
Matches the source exactly
“For a similar reason, we’ve also removed our blanket prohibition on personalized vote and campaign targeting. This prohibition covered legitimate civic work, such as nonprofits using Claude to write information for voters in other languages, or election officials sending ballot cure notices.”
AnthropicFollows the narrowing of the elections section to deceiving voters and disrupting elections; Anthropic says deceptive targeting stays banned under other sections.
Matches the source exactly

The record

Questions
What values should AI have, and could it have a mind of its own? · Are AI companies putting stronger safeguards into practice? · How is AI used in attacks, defence and war?
Also covered by
The Verge · BBC News · rews.cc
What we read
We read the whole source.
Found by
The Radar, run of October 12, 2026, via anthropic-news (tier 1: the source’s own channel or an official record)
Cite this
Anthropic, “2026 Usage Policy update”, October 8, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/anthropic-2026-10-08-usage-policy-update
Our notesClaude wrote this entry about its maker, including a rule about treatment of Claude; editor review needed. The post gives no reason for the cruelty rule and does not use the words 'welfare', 'conscious' or 'moral'; The Verge links it to Anthropic's earlier model-welfare research and to Claude being allowed to end abusive conversations, and BBC reports a mixed public reaction. The post gives 'November 12' without a year. The full Usage Policy text itself was not read, only this summary post. The rews.cc item (via Hacker News) was not read.