What the "screening agent" in a Claude check actually does
The screen is a vendor, not Anthropic, and what it returns is one of three states. Knowing which one you got tells you whether to try again or stop.
Who is asking
When a verification window opens inside Claude, the check is usually run by a third-party identity vendor, not by Anthropic staff. That matters for one practical reason: the decision is automated or semi-automated, and the person you talk to in support cannot override it by asking nicely.
The three outcomes
- Pass. The account continues normally. Limits that were applied before the check may still take a few hours to lift — a pass is not a reset.
- Retry. The most common one. Usually the document photo, the name match or the liveness step. Nothing is decided yet, and another attempt without changing anything produces the same result.
- Fail. Rare, and different from retry: it means the check was completed and the answer was negative. Retrying does not usually reopen it, and what happens to the account and the money is a separate page.
Why "screening" sounds scarier than it is
The word suggests a background investigation. In practice the screen compares a document against a face and against the account, and looks for tampering: a re-photographed document, an edited image, a mismatch between the printed name and the profile. It is a fraud filter, not a character reference.
What tends to trip it
- A photo of a screen or a photocopy instead of the document itself.
- A name written in a different order than the document prints it.
- A connection country that disagrees with the document country, especially when it changes mid-flow.
- Several attempts in a few minutes, which reads as automation.
What to do with each outcome
Retry: change one real cause first — see the first-attempt checklist. Fail: stop retrying and read the consequences page before spending more. Pass but still limited: wait out the limit, do not re-verify.
If you have burned the attempts and cannot tell which of the three you got, that is the case the manual service exists for: somebody looks at the same screen and names the mismatch instead of guessing at it.
How long the screen takes
Automatic checks return in seconds to a few minutes. A screen that does not return is usually not a slow system but a queue: the submission has been routed to a human reviewer, and that reviewer works through a list. Nothing a user does inside the flow moves that list. If the status has not changed in a day, the useful question is not "is it stuck" but "did it come back negative", and a message with the account details to the vendor's support is the way to find out.
What the screen can and cannot see
It sees the document image, the selfie or liveness capture, the name and date of birth on them, and the device and network that submitted them. It does not see your messages, your payment history or your subscription tier — a plan upgrade is not a signal in this system, and neither is a long payment history. That is worth knowing before anyone tells you that buying a higher tier will "help the check go through".
Common questions
Is the screening agent Anthropic staff?
No, it is a third-party identity vendor; the outcome is produced by its system, and support cannot override it.
Does a fail mean a permanent ban?
Not automatically, but a fail is a completed negative check rather than an unfinished one, so retrying rarely reopens it.
Can I skip the screen?
No. It is triggered by the vendor, not chosen by the user, and it either clears or blocks the account features that depend on it.
Guides
All guides →Claude Max 5x vs 20x: which tier you actually need
The difference is not quality — it is how many times you can hit the wall before lunch. A practical way to size your plan.
How to pay for Claude from mainland China (2026)
Four routes, ranked by how often they actually work — and the two that quietly cost you the account.