How to check if ChatGPT, Claude, and Perplexity were hacked

If you suspect that someone has entered your ChatGPT, Claude, or Perplexity account, check active sessions first and then force a fresh login. TechCrunch’s platform-by-platform guide shows that ChatGPT and Claude expose session controls, while Perplexity signs out every session without showing a device list. The practical move is to identify or terminate access before changing credentials.

Definition: A suspicious AI-account session is an active login on ChatGPT, Claude, or Perplexity that you cannot recognize as your own device or browser.

Example: ChatGPT and Claude can show active sessions for inspection, while Perplexity asks you to sign out of all sessions instead of naming one device.

Key takeaway: Inspect sessions where the service allows it; otherwise end every session and sign in again through the account email.

Business impact: Platform-specific session controls let an account owner limit disruption quickly instead of waiting for a provider response or guessing which login is safe.

These checks are account hygiene, not proof of how an intrusion happened. The source recommends unique passwords stored in a password manager and multi-factor authentication where available. ChatGPT and Perplexity offer MFA according to the report; Claude uses an email login link rather than a password. For the broader security picture, AI-agent security control layers show why identity, permissions, tools, and data access should be reviewed together.

What is the fastest first check?

The fastest first check for ChatGPT and Claude is each platform’s own session page, because the TechCrunch guide shows that both services expose a concrete access record instead of requiring you to infer compromise from a strange response. This applies when you still control the account email: open ChatGPT’s or Claude’s security/account settings, look for an unfamiliar session, and terminate it; if you are checking Perplexity, use its global sign-out control because Perplexity does not show a device list.

How do you check ChatGPT active sessions?

ChatGPT lets you inspect and terminate sessions from its browser settings. Select your username in the bottom-left corner, open Settings, choose Security and Login, and click Active Sessions. The resulting list shows where ChatGPT is logged in, so you can log out an individual device you do not recognize or choose Log out all to end every session. OpenAI’s current help documentation also describes the all-devices control in ChatGPT’s Security settings.

If the ChatGPT account needs a new password, sign out first and return to the ChatGPT login page. Enter the account email, choose Forgot password, select Continue, and enter the six-digit code ChatGPT sends by email. The source’s recovery sequence gives the account owner a clean next step after suspicious access: end the sessions, then replace the password rather than leaving the old credential in place.

How do you check Claude active sessions?

Claude places its session controls under Account settings in the browser. Select your username in the bottom-left corner, open Settings, and choose Account to see Active sessions. Hover over a session you do not recognize, select the three vertical dots, and choose Log out or Terminate; Claude also provides a Log out of all devices option. Anthropic’s account guidance confirms that the all-device logout is managed through Settings > Account on the web.

Claude’s recovery path differs from ChatGPT because Claude does not use passwords in the source’s account flow. After logging out, enter the account email and use the login link Anthropic sends there. The operational takeaway is specific: securing the email inbox matters as much as terminating the suspicious Claude session, because the inbox is where the new access link arrives.

What should you do if Perplexity shows no device list?

Perplexity does not show where the account is logged in, so you cannot inspect and remove one suspicious device. If you are concerned about unauthorized access, select your username in the bottom-left corner, open All settings, choose Sign out of all sessions, and confirm. Perplexity’s account-settings documentation describes the same global control for ending sessions on every device and browser.

After the Perplexity sign-out, enter the account email to return. Perplexity sends a unique six-digit code, and its email can also include a Sign in button for direct access. Because Perplexity does not expose a session list, the global sign-out is the evidence-preserving first response: it removes unknown access without asking you to guess which device is safe.

What should you secure after the session check?

For ChatGPT, Claude, and Perplexity, secure the login path after the session check, because the TechCrunch guide recommends unique passwords and MFA where available, while Claude and Perplexity rely on the account email for re-entry. Use a unique password on services that have passwords, store it in a password manager, enable MFA where the platform supports it, and protect the email account that receives recovery codes or login links. For teams using AI inside larger workflows, the AI automation stack is a useful reminder that security follows data and tool connections, not only the model’s sign-in screen.

What this account-security update changes

The practical update is not a new security product; it is a clearer set of recovery paths for three widely used AI platforms. ChatGPT and Claude let an owner review active sessions, while Perplexity requires a global sign-out. Those differences make the correct first action platform-specific, but the sequence stays stable: check or end sessions, regain control through the account email, replace passwords where they exist, and enable the available second factor. The controls are useful only if users actually open them when something looks wrong.