“Your account has been suspended” in Claude: what to do

“Your account has been suspended” means Anthropic has cut off sign-in to your claude.ai account after one of its safeguards fired, and the notice rarely says which one. Whether it ends in a few days or in a permanent ban depends on the cause. Two things are in your hands: filing the appeal properly, and making sure your work does not hang on a single account.

What the message means

Anthropic's help article on safeguards, warnings and appeals, which we read on 10 November 2026, barely uses the word “suspended”. It appears once, in the sentence about accounts that were wrongly suspended or terminated. Everywhere else the article talks about bans and disabled accounts, and about organizations that are paused or on hold. The wording in your email or banner may differ from the help center, so judge the situation by what still works. Can you log in? Does the appeal form open? Do API calls go through?

For banned Free, Pro, Max or Ultra accounts, the article says you can still log in at claude.ai to export or delete your data, though what you can export may be limited depending on the violation. It says nothing about your chat history if the account is reinstated. Export what you cannot afford to lose.

API calls fail differently. The documented errors are 401 authentication_error for a malformed, revoked or expired key, and 403 permission_error for a key without permission to use a resource. The API docs describe no separate error for a suspended account. Save the status code, the response body and the request-id header, because the docs tell you to include the request ID whenever you contact support.

Why Anthropic suspends or bans accounts

Only some of the usual explanations come from Anthropic. The rest are patterns that users describe after the fact.

You cannot rule the reported causes out or in, so work through all of them before you write to support.

Region

The Supported Regions Policy, effective 12 November 2026, says Anthropic's products are available only in the listed countries and regions, and anything unlisted is unsupported. We counted 185 entries. Russia, Belarus, China, Hong Kong, Iran, Syria, Cuba and North Korea are not among them. The policy treats three groups as unsupported: people physically located in an unsupported region, organizations incorporated or headquartered there, and organizations majority-owned or controlled from there.

A sign-up from a supported country does not help if you use the account from an unsupported one. The page does not explain how enforcement changes on 12 November, and it says nothing about independent API gateways. We are not lawyers and draw no conclusions about your case.

VPN and proxies

Many reports tie suspensions to VPN or datacenter IP ranges and to a sudden switch of country between sessions. The usual story: the account worked for months, then one server change ended it. Anthropic has not confirmed a VPN rule in public, so treat it as a pattern users describe, not as a rule. A VPN also leaves your physical location where it was, and the regional policy is written about location. Our longer take is in VPN and Claude bans: what to use instead.

Payment

An expired card, a declined renewal or a bank chargeback comes up often in user reports. The help article does not list billing as a ground for a ban. If your suspension followed a payment problem, fix the payment method and mention it in the appeal.

Unusual activity

Reports also mention volume spikes, dozens of parallel conversations and browser automation. If you run scripts, use an API key instead of the web app. Users say false positives happen, and an appeal is the way to correct one.

How long a suspension lasts

The help article gives no duration. Older guides, including our banned-account guide, quote 24 hours to 14 days. We could not find that figure in the article, so read it as a community estimate. People who describe long cases talk about several weeks, mostly when the reason was unclear and someone had to look at the account by hand. Anthropic says replies are slower than normal and offers no deadline.

Users also report that payment-related suspensions clear faster than ones involving a region or automation flag. That comes from user experience, and Anthropic states nothing of the kind.

How to appeal

1. Search your inbox and spam folder for a message from Anthropic.
2. Write down the exact error text. For API calls, keep the HTTP status, the error body and the request-id header.
3. Log in to claude.ai with the affected account and open the appeal form at claude.ai/restricted. The help article says you must be logged in to reach it.
4. State your account email, what you use Claude for, how often and where you usually connect from. If something changed, say so plainly: you travelled, moved, switched ISP or card, or started running a script.
5. Submit once. We found nothing that says repeat tickets speed things up.

Stay calm in the text and do not embellish. Anthropic can see what the account did, so a story that does not match it will not help.

A warning is a different case from a ban. To dispute a warning, the help article tells you to email the support address printed in it, with your situation and account details.

If you use the API or the Console

The help article describes no appeal procedure for API or Console organizations. It says warnings for API customers depend on ongoing thresholds of violating behavior across the whole API account. If your own account is in good standing but an organization you belong to is paused, the login screen says so, and the article points to a “Request a review” button on that organization. It does not say whether that button covers API or Console organizations.

Keep working while you wait

Waiting weeks for an answer is not an option when production depends on Claude. Claudexia is an API gateway: one key reaches Claude models (Sonnet, Opus, Haiku) and models from other vendors, with no claude.ai login and no Anthropic account. You register with a username and password or with a Google, GitHub or Telegram login, top up from $1 and create a key in the dashboard. A key like that is a separate credential, so your suspension does not switch it off.

Top-ups run through Russian-issued cards (YooKassa, Platega), SBP (the Russian instant bank-transfer system) and crypto (CryptoCloud, CryptoBot in Telegram, or 2328, with USDT, BTC, ETH, TON and others). Which payment route works from which country is covered in the payment guide. Rates are on the pricing page.

Switching an existing Python project takes two lines:

import anthropic

client = anthropic.Anthropic(
    api_key="sk_cdx_YOUR_KEY",
    base_url="https://api.claudexia.tech",
)

message = client.messages.create(
    model="claude-sonnet-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)

Claude Code reads the same settings from environment variables. The ANTHROPIC_DEFAULT_* lines matter: without them the agent asks for its own default models and fails.

export ANTHROPIC_BASE_URL="https://api.claudexia.tech"
export ANTHROPIC_AUTH_TOKEN="sk_cdx_YOUR_KEY"
export ANTHROPIC_DEFAULT_OPUS_MODEL="claude-opus-5-5"
export ANTHROPIC_DEFAULT_SONNET_MODEL="claude-sonnet-5"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="claude-sonnet-5"
claude

The step-by-step version for Windows, macOS and Linux is in Claude Code without an Anthropic account. If you have no key yet, How to get a Claude API key and where to put it covers both routes.

Claudexia does not appeal, restore or replace your claude.ai account, and it does not change Anthropic's regional rules. Whether those rules apply to you depends on where you are and where your organization is registered. A lawyer or Anthropic can answer that. We cannot.

If the suspension turns into a permanent ban, the banned-account guide walks through the types of restriction and the recovery steps.

FAQ

How many suspensions before an account is banned for good?

Anthropic publishes no number. Users report that repeated restrictions for the same reason, especially after a warning from support, raise the risk of a permanent ban, while a single isolated suspension usually ends without lasting consequences.

Does my VPN history matter if I stopped using the VPN?

Nothing we found says an old VPN session is punished after the fact. Reports describe the trigger as the moment the IP jumps to a datacenter range. A stable address lowers the odds of a repeat, but it does not lift a suspension that is already in place.

Can I use the Claude API while claude.ai shows a suspension?

Sometimes. If only the web sign-in is affected, API keys may keep working. When the restriction covers the whole account, API calls fail too. A direct test request tells you which case you are in.

What should the appeal say if the cause is unclear?

Describe your normal use as concretely as you can: tasks, frequency, usual location. Name anything that changed recently, such as IP, device, card or request volume, and explain why. Vague wording is slower to process than specific facts.

Does a complaint on social media speed things up?

Some users report that tagging @AnthropicAI on X got a faster reply. The result is unpredictable and it is not an official channel, so use it as an extra, not a replacement for the form.

Is a suspension the same as a regional block?

The messages can look alike, but the logic differs. A suspension follows a safeguard signal on one account. A regional block is a standing condition for connections from places outside the supported list, and a review will not lift it while you stay there.