How to Check for Unauthorized Logins on ChatGPT, Claude, Perplexity

Update (9 Nov 2026): Claude Fable 5.2 and Mythos 5.2 are out - prices, access and benchmarks are in my Claude Fable 5.2 guide.

👁 118 reads

ChatGPT, Claude, and Perplexity each provide account-security controls that let users spot and remove unauthorized sessions, but the three platforms handle authentication in noticeably different ways. Understanding those differences matters as AI platforms increasingly store conversation history, payment details, and personal data.

The most important first step on any of these services is enabling multi-factor authentication (MFA) where available. ChatGPT and Perplexity both offer MFA. Claude does not - Anthropic’s chatbot skips passwords entirely and instead sends a login link to the user’s registered email address each time they sign in, which means the email account itself becomes the primary security perimeter.

Platform-by-platform session checks

On ChatGPT, finding active sessions requires opening the service in a desktop browser, clicking the username in the bottom-left corner, navigating to “Settings,” then “Security and Login,” and selecting “Active Sessions.” The view lists every device currently logged in. Users can terminate a single suspicious session individually or click “Log out all” to clear everything at once. Changing the password is a separate step: the user must first log out, then use the “Forgot password” flow on the login page, which delivers a six-digit code by email. Entering that code unlocks a prompt to set a new password.

Claude’s session management sits under “Settings” then “Account,” where an “Active sessions” list appears. Hovering over any entry reveals three vertical dots that surface a “Log out” or “Terminate” option for that session alone. A “Log out of all devices” button is also present. Because Claude has no password system, clearing sessions and logging back in via a fresh email link is the only available remediation step.

Perplexity takes the most limited approach: the AI-powered search engine does not display which specific devices are logged in. The only option available under “All settings” is “Sign out of all sessions.” After confirming, the user can log back in by entering their email address and supplying a unique six-digit code sent to that address, or by clicking a direct sign-in link in the same email.

Across all three platforms, security hygiene follows the same baseline: use a unique password stored in a password manager for services that support passwords, and ensure the linked email account is itself protected with MFA and a strong password, since that inbox is the recovery path - or the entire login path - for each of these AI services.

// read moreChatGPT Login: Fix Access Issues and Find Your AccountGuidesPerplexity AI: How to Check Sources and Evaluate ProGuides1Password for Agents: AI Logins Without Sharing Your PasswordNews

Kinetyra