ClaudeMax is an independent reseller, not OpenAI or Anthropic. This page explains the providers' own rules; it is not legal advice, and the official pages linked here always take precedence.
Types of restriction: ChatGPT
| Type | What you see | Typical official trigger | What to do |
|---|---|---|---|
| Usage limit | A model is unavailable until a reset time | Plan allowance used up | Wait or switch model; OpenAI says it does not restrict the account, and Support cannot reset limits |
| Safety refusal or block | The model declines, or a reply is stopped | Usage Policies (for example weapons, sexual content involving minors, malicious cyber activity); classifier-based monitors | Rephrase legitimate requests with context; do not try to get around safeguards |
| Warning | Email or in-product notice about a possible violation | Content or activity flagged by automated systems or human review | Read it and change the behavior; OpenAI says ignoring warnings within the stated time can lead to deactivation |
| Verification request | Verify age, verify an API organization, or verify identity for security work | Age prediction, required age checks in some countries, gated API models or programs, Trusted Access for Cyber | Use only the official prompt or link and your own documents |
| Temporary restriction | Suspicious Activity Alert, some features limited | Unusual sign-ins, usage spikes, many simultaneous logins, suspected compromise or sharing | Change password, turn on 2FA, log out of all devices, then contact Support if it persists |
| Deactivation (ban) | Email saying the account is deactivated; sign-in blocked | Usage Policies or Terms breach, shared account or API keys, repeated violations | Appeal via the link in the email or openai.com/form/appeal |
| Regional block | Service or payment unavailable | Use outside supported countries and territories | Use ChatGPT only where it is supported |
Types of restriction: Claude
| Type | What you see | Typical official trigger | What to do |
|---|---|---|---|
| Usage limit | "Limit reached" with a reset time | Session limit (resets every 5 hours) or weekly limit | Wait for the reset; it is a capacity limit, not an enforcement action |
| Safety block | A reply is blocked, or Claude ends a conversation | Prompt safety filters and classifiers; in rare cases persistently abusive chats | Start a new chat; for blocked security work see the Cyber Verification Program |
| Warning or enhanced filters | Warning email; more sensitive filtering for a while | Prompts flagged as violating the Usage Policy; repeated violations | Change the behavior; dispute a mistaken warning by email |
| Verification request | Identity or business verification, or an age check | Certain capabilities, routine integrity checks, compliance; signals of being under 18 | Complete it through the official flow with your own ID |
| Organization on hold | Login works but an organization is listed as on hold | Unusual activity in that organization | Click Request a review on that organization |
| Suspension or ban | "Your account has been disabled" email; restricted screen | Repeated Usage Policy violations, account created from an unsupported location, Terms violations | Log in with that account and appeal at claude.ai/restricted |
| Regional block | Sign-up or access refused | Location, incorporation or majority ownership in an unsupported region | Use Claude only where it is supported |
Official triggers in detail
OpenAI
The Terms of Use (effective January 1, 2026) set a minimum age of 13 with a parent's or guardian's permission under 18, forbid sharing credentials or making the account available to anyone else, and forbid circumventing rate limits, restrictions or safety mitigations. If a payment cannot be completed, OpenAI may downgrade or suspend access until it is paid. OpenAI may suspend or terminate an account for a breach of the Terms or Usage Policies, to comply with the law, or if use could cause risk or harm.
The Usage Policies (one set for all OpenAI products since October 29, 2025) list what is not allowed: threats and harassment, self-harm promotion, sexual violence, weapons development, malicious cyber activity, fraud and spam, sexual content involving minors, and "circumventing our safeguards". The Pro tiers page adds plainly that sharing account credentials, reselling access and automatically extracting data are prohibited. OpenAI enforces with classifiers, other automated systems and human review, and weighs severity and past violations (transparency page).
Anthropic
The Consumer Terms require you to be 18 or older, forbid sharing login details or making the account available to others, forbid automated access except through an API key, and forbid reselling. Anthropic can suspend or terminate access without notice for a breach, and a subscription terminated for a violation is not refunded. Its help page names three common ban reasons: repeated Usage Policy violations, an account created from an unsupported location, and Terms violations.
The new Usage Policy, published October 8, 2026 and effective November 12, 2026, keeps the "Do Not Abuse Our Platform" rules and widens them. They prohibit creating multiple accounts to avoid detection, circumventing a ban with a different account, giving access in violation of the Supported Regions Policy, using stolen or unauthorized payment methods, and reselling or proxying access to Claude through unauthorized means. The update post also adds a rule against sustained, needless abuse of the models (enforced mainly by letting Claude end such chats) and clarifies that the regional policy covers people physically located in an unsupported region and companies incorporated there or majority-owned from there.
What verification checks, and when it is asked
ChatGPT and the OpenAI API
- Age prediction. ChatGPT may estimate from signals (general topics, time of day, how the account is used, account age) that an account may belong to someone under 18 and switch on the ChatGPT for Teens experience. The account stays active. An adult can review Settings → Account → Age and, if Verify age is offered, confirm with OpenAI's third-party provider (live selfie and/or government ID); the ID or selfie is deleted within 7 days and OpenAI receives only age information (help).
- Required age checks. Where a separate verification notice is shown (OpenAI names Italy, with 60 days), follow its deadline. If an account was deactivated for the minimum-age check, ID review usually takes a few hours (help).
- API Organization Verification. Some API products, models or programs require business verification, identity verification or both, with an original, physical government ID from a supported country. One person can verify only one account or organization, results cannot be manually overridden, and if you are not approved you keep your existing access (help).
- Trusted Access for Cyber. For potentially high-risk defensive security work, OpenAI offers identity- and trust-based access because classifier-based monitors can affect legitimate security work (OpenAI).
Claude
- Identity verification. Rolling out "for a few use cases": when accessing certain capabilities, as part of routine platform integrity checks, or for other safety and compliance measures. You need your physical government photo ID (no photocopies, screenshots or digital IDs) and a phone for a live selfie; organizations also give their legal name, address and registration number. A third-party verification provider holds the images; Anthropic says it does not train on them or share them. You get several attempts, then a help form (help).
- Age assurance. Claude is for adults. If there are signals you may be under 18, the account is disabled and the email links to a third-party age check (selfie estimate, ID or a digital ID app); a pass reinstates the account (help).
- Cyber Verification Program. Generally available models have conservative cyber safeguards, so work such as malware analysis or exploit validation may be interrupted. Security professionals can apply through the official Verification Portal; Anthropic aims to answer within about 7 business days, and the Usage Policy still applies in full (help).
A verification request on its own is not a penalty. Use only the prompt inside the product or the link in an official email, and only your own documents; verifying with someone else's identity, or using an account someone else verified, breaks both providers' rules.
How to appeal, step by step
ChatGPT
- Find the deactivation email and use its appeal link; OpenAI says this routes your case fastest.
- No access to that email? Use the appeal form. No email at all but locked out? Use the Help Center chat (contact Support).
- Include your User ID and Org ID, honest context about your use, the date range if you suspect a compromise, and what you have already done.
- Wait for the reply. A ban stops renewal of eligible self-serve subscriptions; if the ban is lifted, eligible web subscriptions can resume (App Store and Google Play purchases are not restored automatically).
Claude
- Go to claude.ai and log in with the banned account.
- Open the appeal form; it only works while logged in. Describe your use; Anthropic currently warns of longer response times.
- On the same screen, export your data or delete the account if you wish.
- For a mistaken warning, email usersafety@anthropic.com. For a ban after identity verification, the identity page links its own appeal form; for verification problems, the help form.
For scale: Anthropic's Transparency Hub reports 11.4 million banned accounts, 398k appeals and 42k overturns for January–June 2026, so about 11% of appeals succeeded. An appeal is worth filing when you believe the decision is wrong, but it is no guarantee.
What community reports say
These are community reports (not official): individual posts that point to patterns, not confirmed causes.
- Late June 2026, ChatGPT: many paying users on the OpenAI Developer Community described deactivations around June 25 and automated appeal replies; OpenAI support staff asked for case IDs and continued in the support case (thread).
- Plan changes: some of those users suspect plan switches earlier in June as the common factor; OpenAI has not confirmed this (thread). A Claude user on GitHub reported an account disabled right after several upgrades in a few days (March 2026, issue).
- Shared-looking activity: users with repeated Suspicious Activity Alerts quote a support reply describing a temporary automated downgrade for activity resembling account sharing (thread).
- Whole organizations: on Hacker News (April 2026) an employee reported about 70 Claude Team accounts suspended at once without warning (thread).
- Appeal links: a June 2026 GitHub report says the email's appeal link only opened claude.ai; per Anthropic, the form appears only after you log in with the banned account (issue).
Some posts also describe people trying new accounts, other people's verified accounts or hiding their location to get back in. Those are exactly the actions the providers' terms prohibit, and they can turn an appealable case into a permanent one.
How to protect your account, legitimately
- One person, one account. Do not share your login, and do not let others use your account; for a team, use a team plan with separate seats.
- Do not resell or pass on access to your account or API keys.
- Keep payment in good standing. Use a payment method that is yours, and sort out a failed charge through the billing email rather than a chargeback where possible.
- Follow the Usage Policies, especially with agents and automation, and do not use tricks to get around refusals or limits.
- Secure the account: a unique password, two-factor authentication, and logging out of devices you no longer use.
- Use the service where it is supported (see our supported-countries comparison).
- Export regularly: ChatGPT under Settings → Data controls (help), Claude under Settings → Privacy (help).
Both providers say that access from outside their supported lists can be blocked or suspended. Check the current official lists before you sign up, move or travel.