Claude Account Bans, Suspensions and Appeals

The poster guesses Chinese firms use Claude for distillation against Anthropic's rules, so Anthropic mass-bans Chinese accounts and likely sweeps up innocent users.

The poster asks another user whether their account was banned too, saying their own Claude account was suspended in the early hours of the morning.

A user reports that after their Claude account was banned, Capital One issued a full refund of the subscription charge, with screenshots.

An overseas user says they have long chatted with Claude in Chinese without being banned, contrasting with the wave of suspension reports.

The poster shares a six-month Pro account paid via US Google Play on an Oracle US IP with no ban, and asks others for tier, payment method and ban history to compile a survival guide.

A user living abroad asks whether talking to Claude in Chinese could trigger a ban.

A user speculates that Claude account bans may come from repeatedly asking the model how it thinks, which gets misread as extracting chain-of-thought, rather than from network or language issues.

The author maxes out quota every five hours, chats in Chinese and has Claude edit surge/xray configs, yet Claude knows they are in mainland China and has never banned them.

The post relays claims that Claude banned 11.4 million accounts in six months, with only 42k of 398k appeals succeeding, a recovery rate under 0.4%.

Anthropic launched end-customer profiles on its API platform on Oct 20, so a single customer's violation bans only that customer instead of the entire reseller or gateway.

The author sums up Chinese-community talk about Claude bans, including rumors of egress-IP detection, and argues a ban costs far more than the monthly fee.

A quoted user says their account was banned after using Claude Code with ultracode to build a game, suspecting ~13 subagents triggered it.