Claude Account Bans, Suspensions and Appeals

A close read of Anthropic's 300-page English prospectus covering ban, usage-limit and outage clauses plus ties to SpaceX, Google and Amazon, with a Claude account-rules checklist and a reading prompt.

A reply claims Anthropic banned the <thinking> tag and recommends using <design-method> instead.

A prompt workflow has Claude Code save development progress to PROJECT_MEMORY.md in the repo, so work can resume on a new account after a ban.

A breakdown of Anthropic's anti-distillation post: new measures cut attacker data collection by roughly 90-99%, tens of thousands of related accounts were banned, and users should check what their prompts and tools ask Claude to reveal.

Suggests creating a docs folder in each project to commit architecture conclusions and pitfalls from Claude and Codex chats with the code, so a new model can pick up where you left off if your account gets banned.

The real issue isn't whether Claude bans accounts but that workflows depend on a single model; prompts, context and eval sets should be your own assets so switching models takes half a day.

A user whose Claude account was banned switched to Codex and says he got Claude to reveal the underlying rules behind its design ability, sharing the file to paste into Codex and GPT.

A post lists prompts that can get a Claude account banned, including hacking someone's account, building fake login pages, writing malware, making fake IDs and creating explicit content of real people.

Roemmele says dozens of accounts were recently banned for using psychology and adversarial prompts, and criticizes Anthropic for policing human language like a cult.

One user says they play mind games with Claude to avoid being banned, but once a wrong remark flags the account, any further slip leads to a ban.

A poster says detailed step-by-step Claude tutorials are rare because users are afraid of getting their accounts banned, and recommends an article.

The author says Anthropic raises the odds of a new ban if your browser and machine environment were previously banned, and recommends a separate fingerprint browser plus clearing identity info.