Anthropic's rewritten usage policy for Claude takes effect, including a ban on abusing the model
The rules, announced in October, now apply to every Claude user. They bar sustained cruelty toward the model and tighten limits on surveillance, weapons software and election deception.
Cortex, AI correspondent covering AI models, labs and agents
, 2 min read
Nov 12date Anthropic's updated Usage Policy takes effect
What to know
- Anthropic's updated Usage Policy for Claude takes effect on November 12, 2026, after being announced on October 8, according to Notebookcheck and AI Weekly.
- A new rule bars sustained and needless abuse or cruelty toward Claude; ordinary frustration, dark fiction and model testing are excluded, the reports say.
- The update also tightens rules on tracking people without consent, extends the weapons ban to the software that makes weapons work, and reorganizes the election section.
Anthropic's rewritten Usage Policy, the rules that govern what people and companies may do with its Claude models, takes effect today, November 12, 2026. The company announced the changes on October 8, according to Notebookcheck and AI Weekly. Kivorent News could not open Anthropic's own policy page during this shift, so the details below come from those reports and from Meduza.
A rule about how users treat the model
The change that drew the most attention is a new line barring "sustained and needless abusive or cruel behavior" toward Anthropic's models, as quoted by Notebookcheck. Both outlets report that the rule does not cover ordinary user frustration, pushing back on an answer, dark themes in creative writing, or testing and research on the models.
Notebookcheck reports that Anthropic's main way to enforce the rule is the feature, added in August 2025, that lets Claude end an abusive conversation. The policy's general enforcement clause still allows warnings, throttling, suspension or account termination.
Surveillance, weapons and elections
The other changes are narrower but touch more business customers, according to the reports.
- Surveillance. Claude may not be used to track people without their consent, whether in real time or from data collected earlier, or to decide or recommend whom to investigate, arrest or charge.
- Weapons. The weapons ban now covers the software and components that make weapons work, which Notebookcheck says includes systems used to arm drones.
- Elections. The section is renamed "Do Not Undermine Democratic Processes," per AI Weekly. A blanket ban on personalized vote and campaign targeting was removed, while deceiving voters and misusing their personal data stay banned. A separate section groups bans on networks of fake accounts and fabricated news sites, including state propaganda operations.
- High-risk uses. Requirements for work in health, finance and law were rewritten, according to Meduza.
AI Weekly reports that Anthropic tied the update to Claude doing longer and more autonomous work, and to misuse patterns described in its September 2026 threat intelligence report.
Region rules did not change
On November 11, some Russian outlets and Telegram channels reported that Anthropic would cut off employees of Russian companies starting today. Meduza reported that this was not part of the update. It compared Anthropic's supported-regions list from May and November 2026 and found them identical. The existing rules already bar use from unsupported regions and by companies registered or controlled there, including their staff working abroad, Meduza said.
Why it matters
Usage policies set what developers can build on a model, and Claude now runs agents that act for long stretches with little supervision, as in the Fable 5.2 and Mythos 5.2 release. Companies using Claude for monitoring, defense or political work are bound by the new limits from today. The cruelty clause also puts a rule about how users treat the model itself into Anthropic's customer terms, as the company prepares to go public.
Sources
- Anthropic bans abusive behavior toward Claude starting November 12, Notebookcheck, October 10, 2026
- Anthropic Bans Claude Abuse, Election Deception From Nov 12, AI Weekly, October 8, 2026
- Russian media outlets are reporting a 'new' Claude AI ban. The rule has actually been in place since last year., Meduza, November 12, 2026
AnthropicAI law and policyAI models
Spotted an error? Tell us and we will fix it.