How to use Claude anonymously, and what Anthropic still knows on each route

The Morse teamPublished November 6, 202610 min read

To use Claude anonymously, you have to reach it through someone else's account, because Anthropic won't open one for you without a verified phone number. As of November 6, 2026, Anthropic's help center says "We require phone verification for all new users, and there isn't a way to skip this step" (Anthropic), and Claude's incognito chats keep a conversation out of your history but not out of your account. The routes that keep your name off your messages are an app that sends them under its own Anthropic account, such as DuckDuckGo's Duck.ai or Morse's Anonymous mode, and, for developers, Claude on a cloud platform like Amazon Bedrock, where Amazon holds your identity instead of Anthropic.

We read Anthropic's help center, privacy center, privacy policy, API docs, and Commercial Terms on November 6, 2026, along with Amazon's Bedrock docs and DuckDuckGo's Duck.ai terms, and set out below how much Anthropic knows on each route. Anthropic earns credit in two places. Its age check sends Anthropic only a pass or fail (Anthropic), and its API docs tell developers to identify their users with an opaque ID, never a name, email, or phone number.

Can you use Claude without an account?

No, not on claude.ai. Anthropic's sign-in help page describes two ways in, "Continue with Google" or a login link sent to your email, and we found no guest chat on Anthropic's sign-in or help pages. Apple's Hide My Email works with the email option, since Anthropic says those users sign in with their @privaterelay.appleid.com address. The phone check comes next whichever way you sign in.

Can you use Claude without a phone number?

Not with a Claude account of your own. Anthropic's phone verification page, dated August 17, 2026, has every new user verify a number that can take a text, with no skip and no VoIP or Google Voice numbers; we compare that rule with 12 other apps in which AI chatbots need a phone number. The same page has three more rules that decide how anonymous a Claude account can be:

• "We only allow a maximum of three Claude accounts to be verified with the same phone number." A second account made with your number is linked to the first through it.
• "There isn't a way to change your phone number associated with your Claude account once it has been verified." The number you start with stays.
• "Only users physically located in one of our supported locations can create and use Claude accounts."

A mobile number usually leads to a carrier account with a billing name, which makes it a stronger link to you than an email address.

Do the usual privacy tips make a Claude account anonymous?

Mostly not, and some of the guides ranking for this search are wrong about Claude. One, first published in July 2024 and now titled for 2025, says "you can create a Claude account by simply providing an email address, without needing to enter a phone number during sign up" (ExpertBeacon). Anthropic's own rule, quoted above, says every new user verifies a phone number.

A July 2025 guide suggests a "burner email or pseudonym" for the account (Fello AI). The phone number stays on the account whatever email you use. The same guide says deleting your account "triggers complete removal of all associated data." Anthropic's retention page, updated July 1, 2026, says otherwise: chats can outlast a deletion in training pipelines, in records of flagged content, and in feedback you sent, for years in each case. Every period is in our table in is Claude private.

What does Anthropic tie to your Claude account?

Everything you send, plus the details that point to you. Anthropic's privacy policy, effective September 10, 2026, lists:

• At sign-up: "your name, email address, and phone number."
• When you pay: your payment information, "if you choose to purchase access."
• On every visit: your "IP address (including information about the location of the device derived from your IP address)," device identifiers, and what "cookies, scripts, or similar technologies" collect.
• If you're asked to verify: "an image of your government-issued identity document" and "your image in photo or video form," which may include facial geometry templates.

That last item comes from two checks, an age check through Yoti and an identity check through Persona, which we cover in which AI chatbots need a phone number.

The policy also says Anthropic "may share personal data with government authorities, law enforcement, or other third parties" to comply with the law or "an enforceable governmental request." A VPN changes the IP address in that list and nothing else. Your account, number, and card still say who you are.

Is Claude's incognito mode anonymous?

No. Incognito keeps a chat out of your history, and Anthropic still knows which account sent it: the one with your phone number on it. As of November 6, 2026, Anthropic's incognito page says these chats "aren't saved to your chat history" and "are not used for training," and it never describes them as hidden from Anthropic. We list how long Anthropic keeps them in is Claude private.

Incognito doesn't leave your profile behind either. Anthropic says Claude can still use your profile information in these chats, such as custom styles and preferences, though it won't use or add to your memory. Incognito is the right tool against training and a shared or cluttered history, and it leaves your name on every chat.

Can you use Claude anonymously through the API?

Not anonymously, but under different terms. API access starts with a Claude Console account, which Anthropic describes as for "Organizations interested in building with the Claude API," and access "is subject to our Commercial Terms of Service" (Anthropic, August 18, 2026). Usage is "billed through prepaid usage credits" (Anthropic), so Anthropic knows the account and who pays for it. The pages we read don't say whether a Console account needs a phone number.

What changes is what happens to your words: no training, and shorter retention, which we cover in is Claude private. The Commercial Terms also say "Services under these Terms are not for consumer use," so a personal API key is a developer's tool, with a Console organization, keys to manage, and credits to top up.

On a cloud platform, your identity moves to the cloud company. Amazon Bedrock runs each model provider's software in deployment accounts Amazon owns, and Amazon's data protection page says model providers "don't have access to Amazon Bedrock logs or to customer prompts and completions." Anthropic doesn't see you there. Amazon does, through your AWS account and your bill. For Mythos-class models, which Anthropic keeps 30 days on every platform, its covered-models page adds that retained data on Bedrock stays in the cloud provider's environment.

Do third-party apps hide you from Anthropic?

Yes, when the app sends your messages under its own Anthropic account. Anthropic's privacy policy says it "does not apply to content that we process on behalf of customers of our business offerings," so the app's own terms govern what happens to you. Anthropic's API reference shows how little an app is meant to pass along about you: a user ID field that "should be a uuid, hash value, or other opaque identifier," used "to help detect abuse," with the instruction "Do not include any identifying information such as name, email address, or phone number."

DuckDuckGo's Duck.ai works this way, and its terms spell out what it removes and what providers may keep. As of November 6, 2026, its help page lists Claude Haiku 5.5 free, Claude Sonnet 5.5 on Plus, and Claude Opus 5.5 on Pro, "anonymized by us." Its privacy terms, updated October 1, 2026, say metadata such as "your IP address" is "removed before sending Prompts," that providers delete what they receive "at most within 30 days, with limited exceptions for safety and legal compliance," and that they agree to "not using Prompts and Outputs to develop or improve their models." We set it beside Morse in Morse vs Duck.ai.

Any app in the middle learns what Anthropic no longer does: your account, your connection, and possibly your chats. Read its privacy policy for what it keeps.

What Anthropic knows on each route, as of November 6, 2026

Built from Anthropic's, Amazon's, and DuckDuckGo's own pages, linked in the sections above, and Morse's privacy policy.

What can't any route hide?

Your words. On every route but Bedrock, Anthropic receives your message as you typed it, and a message that names your employer, your street, or you identifies you on its own. Anthropic keeps chats used for training "in a de-identified format," but de-identifying a record doesn't edit what you wrote inside it. We go through the gap between de-identified and anonymous in are AI chats anonymized.

How Morse lets you use Claude without Anthropic knowing it's you

Every claude.ai account is pinned to you by a phone number that can't be swapped once it's verified, and the API puts your Console organization and payment on the line instead. The routes around it all put another company's account between you and Anthropic. Morse is a web app with over 30 models, open and closed, Claude among them, and its proxy sends your Claude messages under Morse's own account instead of yours.

You sign in to Morse with a code emailed to you, or with Google, and never make a Claude account, so Anthropic's phone check never applies to you. Your chat history never sits on Morse's servers, so there's no incognito switch to remember. You pick Claude in a chat, under one Morse plan, instead of setting up a Console organization and prepaid credits. As for the app in the middle, Morse's servers don't keep your messages or log what you write, and your Morse account holds your sign-in details, plan, balance, and a usage ledger with no chat content.

Each chat has its own privacy mode:

• Anonymous, to ask Claude with your name, email, and IP address taken off. Pick Claude in an Anonymous chat and our proxy removes your account, email, and IP address, then sends the message under Morse's own account. Anthropic answers without knowing it's you, and may keep the message for a limited time to check for abuse, under its own policy. The proxy takes off who you are and leaves your words as you typed them, so write "my company" rather than its name. Claude Haiku 5.5 is on the Free plan, within 15 messages a day shared with GPT-6 Luna, GLM 5.3 Flash, and more, and Claude Sonnet 5.5 is on Pro, at $19 a month.
• Private, for questions you'd rather no host kept at all. New chats start here. Open-weight models like DeepSeek, Kimi, and GLM, plus xAI's Grok, answer with your identity stripped, and Morse has zero data retention agreements with its inference, search, and other providers. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
• E2EE, for words you'd rather no company read, Anthropic included. Your browser checks the secure enclave's proof that it's genuine before anything leaves your device, then encrypts the message so only that enclave can open it. Morse's servers relay text they can't read. E2EE runs open-weight models, comes with Pro, and depends on the enclave hardware and attestation working as designed. Here's how end-to-end encrypted AI chat works.

Whatever the mode, your chat history is stored in your browser, encrypted, and locked when you sign out. Opening it takes two keys, one that never leaves your device and one our server gives you at sign-in, so a copy of your browser data alone can't be read. Morse doesn't train AI on your chats, sell your data, or show ads. Because we don't store chats, we usually have very little to give authorities. A Temporary chat is kept in memory only and never saved. Clearing your browser or deleting your account deletes your chats, and Morse can't recover them, so keep a copy of anything you need.

Which route fits what you're hiding from?

• Model training: Claude's incognito chats keep a chat out of training inside your account. Morse doesn't train on your chats in any mode, so there's no switch to find. In Anonymous mode, Anthropic's own terms cover what it receives, without knowing it's you.
• Your name on a question to Claude: use an app that sends it under its own account. On Morse, that's Anonymous mode.
• Building a product on Claude: the API, under the Commercial Terms, or Bedrock if you'd rather Amazon than Anthropic held your details.
• The words themselves: every route to Claude hands them to Anthropic or, on Bedrock, to Amazon. For words no company should read, start an E2EE chat on Morse with an open-weight model.

Send a first message from the Morse home page with no account, or start free with an emailed code and no card, and open a Claude chat in Anonymous mode.