AI accounts can reveal signs of unauthorized access

ChatGPT, Claude, and Perplexity offer different ways to review or cut off account access, from active-session lists to full sign-outs.

Just like email, social networks, and messaging apps, AI platforms can also become targets for account break-ins. TechCrunch recently published a guide showing how users can check whether someone may have accessed their accounts on services including ChatGPT, Claude, and Perplexity. The topic is practical, not theoretical: for many people, these tools now hold regular account activity and are used day to day. If an account is compromised, someone else may be able to use the service in the owner’s name or view what has been happening there.

The platforms do offer ways to review or end access, but the process is different on each one. According to IT-PUB News, that matters most when a user is trying to react quickly to a suspicious login.

Passwords and MFA still matter

TechCrunch recommends the standard account-security basics: use unique passwords stored in a password manager and enable multi-factor authentication, or MFA, where the service supports it.

That extra step can make a real difference if a password is stolen. With MFA turned on, a login attempt may still be blocked unless the attacker also has the second factor. Among the platforms covered here, ChatGPT and Perplexity support MFA. Claude does not use passwords at all, relying instead on a login link sent to the user’s email address.

Those differences shape what users can actually do when they want to check for suspicious access. ChatGPT and Claude let people inspect active sessions directly. Perplexity does not show where a user is logged in, so the options are narrower.

ChatGPT shows active sessions in account settings

To check whether a ChatGPT account may have been accessed by someone else, users need to open the service in a browser on a computer. Then they should click their username in the bottom-left corner, open “Settings,” then “Security and Login,” and select “Active Sessions.”

That page shows where the account is currently logged in. If a user sees a device they do not recognize, they can log out of that single session or choose “Log out all.”

If a password change is needed, ChatGPT requires the user to log out first. After that, they can return to the login page, click “Log in,” enter their email address, choose “Forgot password,” and continue through the reset process. ChatGPT then sends a six-digit code by email. Once the code is entered, the user can set a new password.

The service also includes a “reset your password” link in the email, which leads to the official instructions.

Claude relies on email login links

Claude handles account access differently because it does not use passwords. Instead, it sends a login link to the user’s email address.

To check for suspicious access, users should open Claude in a browser on a computer, click their username in the bottom-left corner, then go to “Settings” and “Account.” There, they will find “Active sessions.”

If an unfamiliar session appears, users can hover over it, click the three vertical dots on the right, and choose “Log out” or “Terminate.” Claude also offers a “Log out of all devices” option.

After that, logging back in is done through email. The user enters their email address and receives a link to sign in. Since Claude does not allow passwords, there is no password to change.

Perplexity offers only a full sign-out

Perplexity handles session checks differently again. The AI-powered search engine does not show users where they are logged in, which means they cannot review a list of active devices the way they can on ChatGPT or Claude.

If there is concern that an account may have been compromised, users can still force a sign-out from all sessions. To do that, they should open Perplexity in a browser, click their username in the bottom-left corner, go to “All settings,” then select “Sign out of all sessions” and confirm the action.

After signing out, users can log back in by entering their email address. Perplexity then sends a unique six-digit code to the user’s email. The code can be entered on the website, or the user can click the “Sign in” button in the email to log in directly.

AI services now need the same account checks as other apps

The guide points to a broader shift: AI platforms are no longer niche tools, but regular online accounts exposed to many of the same risks as other services. That makes basic account hygiene important for people using them for work, personal projects, or everyday searches.

The three platforms also do not treat account access the same way. ChatGPT and Claude let users inspect sessions and remove specific devices, while Perplexity offers only a full sign-out. In practice, that can affect how precisely and how quickly someone can respond when something looks off.

For users who rely on these services, the practical takeaway is straightforward. Check active sessions where the platform allows it, sign out devices you do not recognize, and reset access if needed. The steps vary, but the goal does not: keep control of the account before someone else does.