Your AI accounts are now valuable enough to steal. Hackers target ChatGPT, Claude, and Perplexity just like they go after email or bank logins, and the consequences can be worse than you think. Someone inside your ChatGPT account can read your conversation history, which might include sensitive work documents, personal details, or proprietary code you pasted for debugging. That makes these platforms a goldmine for identity thieves and corporate spies alike.
TechCrunch walked through the security settings on all three platforms, and the process varies significantly. ChatGPT makes it straightforward: open settings on desktop, navigate to Security and Login, then click Active Sessions. You’ll see every device currently logged in. If something looks off, you can boot that single device or nuke every session at once. Password changes require logging out entirely, then clicking ‘Forgot password’ to receive a six-digit code by email. Anthropic’s Claude is simpler in one respect but frustrating in another. The platform doesn’t use passwords at all — it sends a login link to your email — so there’s no password to steal or reset. However, Claude also doesn’t offer multi-factor authentication, which means anyone with access to your email inbox effectively owns your Claude account. That’s a glaring security gap for a platform handling increasingly sensitive conversations.
Perplexity takes the most opaque approach. Unlike its competitors, the AI search engine won’t show you where you’re logged in. There’s no active sessions list to review. Your only option is to sign out of all sessions and log back in, which requires a six-digit code sent to your email. It’s a blunt instrument, but it gets the job done if you suspect a breach.
Here’s the uncomfortable truth: two of these three platforms still lack basic security hygiene. Claude’s email-only login without MFA is a serious liability, and Perplexity’s refusal to show active sessions means users are flying blind. Compare that to Google or Microsoft, which have offered granular session controls and robust MFA for over a decade. AI companies are moving fast to capture market share, but security fundamentals are lagging behind. Until that changes, your best defense remains the boring stuff: unique passwords in a password manager, MFA wherever it’s offered, and a habit of checking active sessions before something goes wrong rather than after.
