How to tell if your AI platforms’ accounts have been hacked

Details

As AI platforms become daily-use tools, account security is a growing concern for individuals and enterprises alike. TechCrunch's guide walks through session management UIs for the three most widely-used AI chatbots. For ChatGPT, users can view active sessions under Settings → Security and Login → Active Sessions, revoke individual devices, and reset passwords via email verification. Perplexity lacks per-device session visibility but offers a blanket sign-out-all-sessions option.\n\nClaude's situation is structurally different: Anthropic uses email magic links exclusively — no passwords, no MFA support. This means users cannot change a compromised password; they can only terminate active sessions. While this eliminates password-based phishing attacks, it also means anyone with email inbox access has full Claude account access.\n\nFor security practitioners, Claude's absence of MFA is a notable gap as Anthropic pushes enterprise adoption. The practical advice — check active sessions regularly, enable MFA where offered (ChatGPT and Perplexity support it), use strong unique passwords — applies directly to AI platform hygiene that most organizations have not yet added to their security runbooks.