Claude Code 403 "Your Account Can't Receive Replies": Next Step

Failed to authenticate. API Error: 403 Your account can't receive replies from Claude right now means Anthropic's server has stopped answering this account until it completes identity verification in the Anthropic Verification Portal. It is not a Claude Code bug, a network fault or an outage, so retrying, waiting or running /login again with the same account will not clear it. In a reproduction against a local mock server on October 30, 2026, Claude Code 2.1.312 ended the run after a single request, with no retries.

Before you click anything, run claude auth status --text to see which account Claude Code is signed in with; in Claude Code 2.1.312, claude auth status --help on October 30, 2026 listed --text for readable output and --json as the default, and /status inside a session shows the same login. Then verify at portal.anthropic.com with that account only if two things are true: you are physically in a location Claude supports, and you have a valid physical government photo ID, such as a passport, driver's license or national ID card, in your hand. If both hold, verify now; that is the case the check is built for.

If you are physically somewhere Claude doesn't support, such as mainland China, Russia or Hong Kong, verifying does not make that use allowed, and several users reported their accounts suspended right after verifying on October 29, 2026. Your local files are untouched either way; what stops is Claude's replies. Don't buy an "unlock" or borrow someone's ID, because both break Anthropic's rules and put the account at further risk.

What "Your account can't receive replies from Claude right now" means

This is an account-level identity check, often called KYC by users, that Anthropic sends from its server; Claude Code only prints the response it received. The full line looks like this, with your organization's ID at the end of the link:

Failed to authenticate. API Error: 403 Your account can't receive replies from Claude right now. You can find the next step in the Anthropic Verification Portal: https://portal.anthropic.com/verify?org=...

A byte search of the shipped Claude Code 2.1.312 binary on October 30, 2026 found no copy of "receive replies" anywhere in it, so the client has no built-in message with this wording and is passing the server's text through. The "Failed to authenticate" prefix is the label Claude Code puts on both 401 and 403 responses (it printed it for each in the test below), so it does not mean your key or login is broken. As of October 30, 2026, the Claude Code error reference doesn't list this message at all.

Anthropic's identity verification help article says the company is rolling out verification "for a few use cases," including routine platform integrity checks, and that the check is run by its partner Persona. The same wave shows up on claude.ai as "Replies from Claude are unavailable on your account. You can complete the verification now in the Anthropic Verification Portal," or as a "Complete verification to continue" banner; the Claude KYC Verification: Phone, ID, Payment, Age, and Bans guide covers those claude.ai screens and every other verification flow.

Retrying, waiting or /login won't fix this 403

Claude Code gives up on this response at once, and that is deliberate: it does not retry a 403 from Anthropic, so the same account will get the same answer until the verification state changes. Anthropic's Claude Code documentation on automatic retries says Claude Code retries server errors, overloads, timeouts, dropped connections and temporary 429s up to 10 times, but retries a 401 or 403 only when an apiKeyHelper script supplies the credential.

A local test on October 30, 2026 shows the difference. A mock server on the same machine answered every request with this exact message and HTTP 403; it was not Anthropic's server and no real account was involved, and the 403 status was taken from a Claude Code user's post on X on October 29, 2026. Claude Code 2.1.312 in claude -p mode, with either an API key or a bearer token, printed the same line that user posted, sent one request, did not retry and exited with code 1 within 1–3 seconds. When the same mock returned a 401 instead, Claude Code retried 10 times with growing gaps and gave up after about 3 minutes. So if you see this message immediately, there is nothing transient to wait out.

Signing in again doesn't help either, because the check is attached to the account rather than to the saved login, and /login with the same account brings you straight back to it. If your 403 has different wording, such as a proxy, region or gateway refusal, start with Claude Code 403, 503, and 529 Errors: Find the Source, Then Fix; an "unable to authenticate" message is a separate, often temporary login problem covered in Claude (Temporarily) Unable to Authenticate: Outage or Local Fix?.

Should you verify? It depends on your location and your ID

Verify now if you are physically in a supported location and have a valid physical photo ID in hand; otherwise the right move depends on which of those two conditions you fail. Anthropic's list of supported locations says only users physically located in a supported location can create and use accounts, and the identity verification article lists which IDs it accepts.

Reports from people who already verified are mixed. In one X thread started on October 28, 2026 by a user asking how to rescue a years-old account, about 15 replies arrived within a day: two said they verified and were then suspended, one of them after choosing China as the country and uploading a passport (the Claude Code user whose October 29 post on X is cited above said the same, so three people in all), one said a passport didn't work, and two said a Chinese passport can pass. These are unverified self-reports, mostly from users in mainland China, which is not a supported location, so they say more about that rule than about verification in general.

How to verify through the Anthropic Verification Portal

The order matters because the portal verifies an account, not a computer, so confirm the account first and then complete the check in one sitting.

1. In a terminal, run claude auth status --text, or type /status inside a Claude Code session, and note the email and organization it shows.
2. Open the link printed in the error, or go to portal.anthropic.com directly, and sign in with that same account. Anthropic says verification should start only from a prompt inside Claude, your settings, an email from Anthropic or portal.anthropic.com, so ignore links sent by strangers. The "Cyber Verification Program" on the same portal is an unrelated opt-in for security professionals.
3. Have the physical ID and a phone camera ready, and expect a possible live selfie. The ID must be legible, undamaged and unexpired; Persona holds the images, and Anthropic says they aren't used for training.
4. If the flow rejects your attempts, stop after the attempts it allows and use the help form linked from the identity verification help article instead of retrying with other documents.
5. Return to Claude Code and run your last command again to see which state the account is now in.

What Claude Code shows after you verify

After verification, the message Claude Code prints tells you which state the account is in, and each state has a different next step. The "on hold" text below is built into Claude Code 2.1.312 and documented in the error reference. Anthropic hasn't published how long portal review takes, so the first row is what a cleared account should look like rather than a confirmed screen.

Anthropic's documentation describes "on hold" as the Claude account behind your login having been suspended, and signing in again with the same account doesn't clear it. If that is where you land, the full appeal and refund rules are in Claude Code Max Account Banned After Recharge? Appeal and Refund. Two facts matter right away: you appeal by logging in at claude.ai with the suspended account and using the form at claude.ai/restricted, and suspended Free, Pro and Max users can still log in to export their data or delete the account, according to Anthropic's safeguards, warnings and appeals page.

Stop scripts and CI from looping on the 403

In claude -p scripts, check the api_error_status field rather than subtype, because a failed run still reports "subtype":"success". In the October 30, 2026 mock test, --output-format json returned these fields alongside the error text in result, and the process exited with code 1:

{"subtype":"success","is_error":true,"api_error_status":403,"terminal_reason":"api_error"}

A wrapper that treats a 403 as final, instead of feeding it into a retry loop, can look like this:

out=$(claude -p "$PROMPT" --output-format json)
if [ "$(printf '%s' "$out" | jq -r '.api_error_status // empty')" = "403" ]; then
  echo "Claude account needs verification; stopping without retry" >&2
  exit 2
fi

Can you keep working while the account waits?

You can keep coding with another credential only if that credential belongs to an account that isn't affected, and Anthropic hasn't said whether an API key works while your Claude account waits for verification. Anthropic's Claude Code documentation suggests switching to another Claude account or an unaffected API key with /login or ANTHROPIC_API_KEY while an account is on hold, and its help center says a Claude account and a Console account can share an email but operate independently. If the key's own organization gets this message, it fails exactly the same way: the mock test above produced the identical line with an API key. Claude Code API Billing vs Subscription: Key, Credits, or Pro/Max explains which credential Claude Code actually uses when both are set.

Claude Code can also point at a different Anthropic-compatible endpoint through ANTHROPIC_BASE_URL, either for another model, as in GLM-5.3-Flash in Claude Code: Setup That Actually Runs Flash, or for Claude models through a separate API service such as laozhang.ai, whose docs as of October 30, 2026 cap output at 32,768 tokens and don't support Anthropic's web_search, web_fetch or code_execution tools, Batches or Files. Any such service has its own terms, and none of these options restores your Anthropic account or makes using Anthropic's service from an unsupported location allowed.

Disclosure: laozhang.ai is our own service. Its prices are platform pricing, not an independent comparison.

Unlock sellers, borrowed IDs and new accounts break Anthropic's rules

Paying someone to "unlock" the account, verifying with another person's ID or opening a fresh account to get around a suspension all violate Anthropic's Usage Policy, so none of them is a safe way back. Offers already exist: as of October 30, 2026, a Russian-language Threads post quoted this exact message and advertised unlocking through a Telegram contact, which is not part of Anthropic's process. Anthropic's Usage Policy prohibits using multiple accounts to avoid detection, circumventing a ban through a different account, bypassing sign-up or log-in restrictions and accessing Claude in violation of the Supported Regions Policy, and the Consumer Terms forbid sharing your account or credentials. Handing a stranger your ID images or login also gives them something to misuse long after the account is settled.

Why did an old Claude account get the verification request now?

As of October 30, 2026, Anthropic hasn't explained why this request reached accounts that had been in use for months or years on October 28 and 29, 2026. The help article only says verification can appear when you access certain capabilities, during routine integrity checks or for other safety and compliance reasons, and it doesn't mention Claude Code. User reports from those two days describe many long-standing accounts hit on the same day, with one user writing "a whole wave today," but they come mostly from users in mainland China, so they don't show who else is being asked. Anthropic also hasn't said whether a Console API key in a separate organization is affected, how long portal review takes, or whether passing verification guarantees continued access.