Which AI chatbots need a phone number? What 13 apps ask for to sign up

The Morse teamPublished October 25, 202610 min read

Most AI chatbots work without a phone number now. Of the 13 AI chat apps we checked on October 25, 2026, Claude is the only one whose own help pages require a phone number from every new user, and seven (ChatGPT, Gemini, Copilot, Perplexity, Duck.ai, Proton's Lumo, and Venice) let you chat without an account at all. Age checks have grown instead. ChatGPT, Claude, Character.AI, and Google, which runs Gemini, can each end up asking for a selfie or a government ID.

We read each company's own help pages, sign-in page, privacy policy, or terms on October 25, 2026, and put the answers in one table, every cell linked to its source. Many results for this search are older. A 2023 guide that still ranks says ChatGPT "requires users to provide their phone numbers" and suggests Google Bard instead, and one from July 2025 calls a phone number "one of the main requirements for using ChatGPT."

What each AI chatbot asks for, as of October 25, 2026

"None described" means the pages we read don't mention one. It doesn't prove an app never asks.

Does ChatGPT need a phone number?

Not for most people, as of October 2026. OpenAI's phone verification page says "Phone verification is not generally required to create an OpenAI account or use ChatGPT" (OpenAI). Two places still ask: Codex in the ChatGPT desktop app, and the first API key you create on OpenAI's developer platform.

OpenAI's phone-only sign-up page confirms the rule changed, in an example about accounts made "from the older days when phone verification was required at signup" (OpenAI). In some countries you can now sign up with only a phone number, but that account can't upgrade to Plus, Pro, or Business until you verify an email.

A number you add stays. OpenAI says "You cannot unlink a phone number from an existing OpenAI account" (OpenAI). Our page on anonymous ChatGPT covers what signing out hides from OpenAI and what it doesn't.

Does Claude require a phone number?

Yes, for every new account. Anthropic's help center says "We require phone verification for all new users, and there isn't a way to skip this step," and it rejects VoIP numbers, Google Voice, numbers made with apps, and landlines (Anthropic, dated August 17, 2026). Its privacy policy lists your "name, email address, and phone number" as what it collects when you sign up (Anthropic).

Claude has two more checks, each set off by a trigger:

• Age. Claude is for adults only, and when Anthropic detects "signals that you may be under 18," it disables the account until you pass a check with Yoti: a selfie for facial age estimation, a photo of your ID, or Yoti's Digital ID app. Anthropic says it receives only a pass or fail result (Anthropic, May 18, 2026).
• Identity. Anthropic's help center (August 11, 2026) says some capabilities and "routine platform integrity checks" can ask for a physical government photo ID and a live selfie through Persona. Anthropic can view those records through Persona when it needs to, for example to review an appeal (Anthropic).

We cover what Anthropic keeps and trains on in is Claude private.

Which AI chatbots work without an account?

Seven of the 13 let you start without signing up, with limits:

• ChatGPT supports "one conversation at a time" when you're signed out (OpenAI), and training stays on unless you turn off "Improve the model for everyone" in that browser's settings (OpenAI).
• Gemini lets you "use some features" on gemini.google.com without signing in (Google). Google says signed-out data, "like your IP address and browser type," falls under its general privacy policy and "is not associated with your Google account" (Google).
• Copilot works signed out, and signing in adds "chat history, image creation, longer conversations, voice interactions" (Microsoft).
• Perplexity keeps "logged-out threads" for 14 days, after which they "are not recoverable" (Perplexity, March 3, 2026).
• Duck.ai is "free to use and anonymized by us, with no account required" (DuckDuckGo), within a daily limit. Its paid subscription doesn't need your email either; you add one only to use the plan on a second device (DuckDuckGo).
• Lumo offers guest access "with no need to log in or create a Proton account," including over Tor, and a guest conversation "is erased at the end of each session" (Proton).
• Venice says in its App Store listing, "Start creating in seconds for free. No sign-up required" (Venice).

Which AI chatbots ask for a selfie or ID?

ChatGPT, Claude, Character.AI, and Google, and none of them asks at sign-up. Each asks only when its system guesses you might be under 18, or for Claude, when you reach certain capabilities.

• ChatGPT predicts your age from signals such as "general topics you talk about, the times of day you use ChatGPT," and how long your account has existed. If it guesses under 18, you get the teen experience until you verify with Persona by live selfie, government ID, or both. Persona deletes them within 7 days, and OpenAI receives your date of birth, not the images (OpenAI).
• Character.AI runs age assurance on every account, from "login info, activity on the platform, and some signals from third parties." If it flags you as under 18, you can accept a reading-only account or take a Persona selfie, and ID comes up only if Persona isn't sure from the selfie (Character.AI).
• Google estimates age from signals "such as the types of information a user has searched for or the categories of videos they've watched on YouTube," and an adult it misjudges can upload a government ID or a selfie (Google, July 30, 2025).

Can you use a VoIP or Google Voice number instead?

Not for the checks that remain. Claude's sign-up and OpenAI's Codex and API checks both turn away VoIP and Google Voice numbers. OpenAI says it also rejects landlines and premium numbers because "these types of numbers are often associated with higher instances of fraud or abuse" (OpenAI). The burner-number advice in older guides was written for a ChatGPT check that most people no longer see.

Other common claims are out of date too. Character.AI dropped Discord, Facebook, and email-and-password logins on July 30, 2024, and now signs you in with Google, Apple, or an emailed link (Character.AI). Some guides say DeepSeek asks for a Chinese +86 number, but DeepSeek's English-language terms say to register "using your Email or third-party account" (DeepSeek).

Why does giving an AI chatbot your phone number matter?

An email address can be made up in a minute. A mobile number usually comes with a carrier account, a billing name, and years of history, so it ties an AI account to a real person in three ways:

• It's what a subpoena asks for. US federal law lets the government get a service's subscriber records, including name, address, "telephone or instrument number," and how you paid, "including any credit card or bank account number," with an administrative, grand jury, or trial subpoena, no warrant needed (18 U.S.C. § 2703(c)(2)). A phone number on an AI account points investigators to a carrier that keeps its own records of who pays for the line. This isn't legal advice. We go through how chats themselves reach courts in can AI chats be subpoenaed.
• Numbers given for security have been used for ads. In 2022 Twitter agreed to pay a $150 million penalty to settle FTC charges that more than 140 million users gave it phone numbers or emails to secure their accounts, and that it let advertisers use them to target ads (FTC, May 25, 2022). Google says openly that its "Better ads & Google services" setting lets your phone number be used across its services to show you more relevant ads, and you can turn it off (Google).
• It becomes the key to your account. Phone numbers reset passwords, and a number can be stolen. The FCC adopted rules on November 15, 2023 against SIM swapping, where a scammer "convinces a victim's wireless carrier to transfer the victim's service" to their own phone (FCC). Whoever holds the number gets the reset codes for any account that sends them by text, AI accounts included.

Deleting the account doesn't always remove the number right away. Mistral keeps the phone number or email "you provided to verify or change your account password" for one year after you delete your account (Mistral). Proton handles this best of the apps we checked. For human verification it keeps "only a cryptographic hash" of your number or email, enough to spot reuse without storing the number itself (Proton).

How Morse keeps your AI chats from being tied to you

In most rows of the table, the company that makes the model holds an account it can tie to you, by phone number, email, or an ID check, and keeps your chats under it. Morse is a web app that brings over 30 models, open and closed, into one account, built so the model's maker never gets your account and your chats don't sit on our servers.

You can start without any account at all: a message sent from the Morse home page opens a guest chat that's kept in the tab's memory only and never saved. When you want to keep chats, you sign in with a code emailed to you, with no password, or with Google. Morse is for adults, and you confirm you're 18 or older. The Free plan needs no card.

Each chat then has a privacy mode:

• Anonymous, to use Claude, GPT, or Gemini without an account of your own there. Morse's proxy removes your account, email, and IP address and sends the message under Morse's own account. There's no Anthropic account to set up, and the provider sees the question without knowing it's you. It may keep the message for a limited time to check for abuse, under its own policy, so leave your name and other details out of what you type.
• Private, for chats no host should keep. It's the default for a new chat. Open-weight models like DeepSeek, Kimi, and GLM, plus Grok, run with your identity stripped, and Morse has zero data retention agreements with its inference, search, and other providers, so they've agreed to keep nothing.
• E2EE, for words no one should read. Your browser checks the secure enclave's proof that it's genuine, then encrypts the message so only that enclave can open it, and Morse's servers relay text they can't read. It runs open-weight models, depends on the enclave hardware and attestation working as designed, and comes with Pro, at $19 a month. Here's how end-to-end encrypted AI chat works.

In every mode, your chat history is stored only in your browser, encrypted, and Morse's servers don't keep messages or log message content. Your Morse account holds your sign-in details, plan, balance, and a usage ledger with no chat content. Because we don't store chats, we usually have very little to give authorities. Morse doesn't train AI on your chats, sell your data, or show ads (Morse privacy policy). Clearing your browser or deleting your account deletes your chats, and Morse can't recover them, so save a copy of anything you need.

Send a first message from the Morse home page with no account, or start free with an emailed code and no card.