airescues / chatgpt / ChatGPT hub

ChatGPT by OpenAI

Someone else is in your ChatGPT account: lock it down in this order

Order matters here: change the password first, log out of every session, and only then turn on MFA.

Shut the intruder out

  • Change the password first: all current sessions are logged out within 30 minutes.
  • Sign in with Google or Microsoft? Reset that account’s password instead.
  • In Active sessions, choose Log out all, then Log out of all devices.
  • Change the password and log out of all sessions before enabling MFA.
  • Use the API? Delete your API keys and review usage for unexpected activity.
An illustrative bar chart of signed-in sessions on a ChatGPT account: two normal sessions, then extra sessions after someone else gets in; after the password change every current session drops away within 30 minutes, leaving only you, signed in again.
Changing the password logs out all current sessions within 30 minutes; you then sign in again. Illustrative numbers.

Read what the intruder did

  1. Security history sits under Settings → Security and login.
  2. A single session can be ended on its own in Active sessions.
  3. Active sessions isn’t available for accounts linked to an organisation’s SSO.
  4. It doesn’t list third-party app or Codex CLI sessions.

Tell OpenAI

  • Use the Unauthorized Activity webform and choose «Report unauthorized activity involving my account».
  • The fraud webform link points to openai.com/form/appeal/.
  • If OpenAI suspects a compromise, it may suspend the account temporarily to protect you.
  • Write to Support from the chat bubble at the bottom-right of help.openai.com.

OpenAI publishes no separate procedure for handing a hijacked account back beyond Support and these webforms.

Turning on MFA after the clean-up

  1. MFA on its own does not log out other devices or sessions — that is why the password change and the log-out come first.
  2. Turn it on under Settings → Security and login → Multi-factor authentication (MFA).
  3. Pick an authenticator app, a push approval, an SMS or WhatsApp code, or a passkey.
  4. It then guards ChatGPT and the API Platform together.
  5. Add a backup method as well: the one-time recovery through Support works only once.

Charges made while someone else was in

  1. Contact your bank straight away if a charge looks unauthorized.
  2. Unauthorized charges are among the exceptions to OpenAI’s general no-refund rule.
  3. OpenAI validates such charges before any refund is issued.
  4. While a card dispute is active, OpenAI won’t issue a separate refund.
  5. Can’t sign in to cancel a plan you didn’t buy? Support can cancel with the account email plus any two billing details — never the complete card number.

Prompts and alerts you didn’t start

  1. A push approval you didn’t start: tap «No, deny access».
  2. Genuine login-code emails come from [email protected] or [email protected].
  3. Suspicious Activity Alerts are often cautionary; sharing the account can trigger them.
  4. Be cautious with emails that ask for credentials or send you to pages that want account details.
recovery is not for sale A «recovery service» that wants your password or the code from your phone is the next intruder, not a rescuer. OpenAI says it does not request passwords or verification codes by email, text or chat, and its messages come from official OpenAI domains.

Next for ChatGPT

Sources — OpenAI help pages, checked October 2026

AIRescues explains the official ways back into your own AI account. We are not ChatGPT, Claude, Gemini, Grok or DeepSeek and cannot open any account for you.

Facts are checked against each platform’s help pages; every guide shows the month it was checked.

How we check About Responsible use