How to tell if your AI accounts have been hacked

Hackers can target and break into accounts on AI platforms just as they can on any other online service. TechCrunch recommends the basics everywhere — a unique password stored in a password manager and multi-factor authentication (MFA), so a stolen password alone isn’t enough to log in. ChatGPT and Perplexity both offer MFA; Claude doesn’t, because instead of a password, Anthropic’s chatbot sends a login link to your email address.

All three platforms let you check for a suspicious device logged into your account, though the process differs. On ChatGPT, click your username in the bottom-left corner, then go to Settings > Security and Login > Active Sessions to see where you’re logged in; you can log out a single unrecognized device or click “Log out all.” To reset your password, log out, click “Log in,” enter your email, select “Forgot password,” and ChatGPT will email you a six-digit code to enter before setting a new one.

On Claude, click your username, then Settings > Account to see your active sessions; hover over an unrecognized one, click the three dots, and choose “Log out” or “Terminate,” or log out of all devices at once. Since Claude has no passwords to change, you simply log back in with your email address and a link Anthropic sends you. Perplexity doesn’t show where you’re logged in at all — instead, click your username, then “All settings,” and select “Sign out of all sessions” to force every device to reauthenticate; you’ll log back in with your email and a six-digit code Perplexity sends you.

Why it matters

As AI chatbots become repositories of personal and professional context, keeping their accounts secure is no longer an afterthought — and the platforms’ account-security tools vary enough that it’s worth knowing exactly how each one works before you need it.