Securing Your AI Platform Accounts: What to Do if Hacked

Artificial intelligence platforms have become essential tools for millions of users worldwide, but they come with the same security vulnerabilities as any other online service. Your ChatGPT, Claude, and Perplexity accounts can be compromised by attackers who gain unauthorized access through stolen credentials, phishing, or weak security practices. Understanding how to detect and respond to a breach is critical for protecting your data and preventing misuse of your account.

Why AI Accounts Attract Hackers

AI platform accounts hold significant value to attackers. These accounts often contain conversation histories, saved preferences, API keys for developers, and access to advanced computational resources. For some users, a compromised AI account could expose sensitive work, creative projects, or personal information shared during conversations. The platforms themselves have become mainstream enough that they’re now on the radar of criminals who once focused solely on email and banking services.

Three Platforms, Three Different Security Models

Read also: Flock’s Abuse Detection Tool Faces Skepticism Over Missing Details · Uber Freight confirms data breach by Helix extortion gang · Pattern-Generating AI Defeats Surveillance Camera Detection in First Real-World Test

The major AI chatbots handle authentication differently, which shapes how you should approach security on each platform. Understanding these distinctions is the first step toward protecting yourself.

Authentication Methods Vary Significantly

ChatGPT and Perplexity both use traditional passwords combined with optional multi-factor authentication. Claude takes a different approach: Anthropic’s chatbot abandons passwords entirely in favor of email-based authentication. When you log into Claude, the platform sends you a unique login link to your email address rather than asking for credentials. This eliminates an entire class of vulnerabilities—there’s no password database to breach—but it does rely on the security of your email account. ChatGPT and Perplexity offer multi-factor authentication as an additional security layer, while Claude does not because the email-link system is considered sufficient protection.

Session Visibility Differs

All three platforms allow you to view where you’re currently logged in, though with varying levels of detail. ChatGPT provides the most transparent view of your active sessions, showing each device and location. Claude similarly displays your active sessions with the ability to review and terminate them individually. Perplexity, however, takes a less detailed approach—it does not show you where you are logged in, leaving you with fewer options for granular control. Instead, Perplexity offers a nuclear option: sign out of all sessions at once.

Checking ChatGPT for Unauthorized Access

ChatGPT makes it relatively straightforward to audit your account activity and remove unauthorized sessions.

Locating Your Active Sessions

Open ChatGPT in your browser and locate your username in the bottom left corner. Click on it to open a menu, then navigate to Settings. From there, select Security and Login, and finally click on Active Sessions. This screen displays every device currently logged into your ChatGPT account, showing where each session is active. Review this list carefully and look for any devices, locations, or login times that seem unfamiliar to you.

Removing Unauthorized Devices

If you spot a suspicious session, you have two options. You can log out of that single device by selecting it and confirming the logout, which leaves your other sessions intact. Alternatively, if you suspect a more serious breach or want to be certain no attacker remains in your account, click Log out all to end every active session at once. This forces you and any unauthorized users to log back in.

Resetting Your Password

After logging out, you’ll need to create a new password to reclaim your account. Visit ChatGPT’s login page and click Log in in the bottom-left corner. Enter your email address, then click Forgot password followed by Continue. ChatGPT will send you an email containing a six-digit code. Enter this code on the login page, click Continue, and then create a new, strong password. Alternatively, you can click the reset link in the email ChatGPT sends to complete the process through a direct verification flow.

Protecting Your Claude Account

Claude’s passwordless design streamlines the security recovery process, though it requires understanding how the email-based system works.

Finding Your Active Sessions

Log into Claude in your browser. Click your username in the bottom-left corner, then select Settings and Account. Under Active sessions, you’ll see every current login associated with your account. Each session displays enough information for you to identify whether it’s legitimate or suspicious.

Terminating Compromised Sessions

To remove an unrecognized session, hover over it and three vertical dots will appear on the right side. Click these dots and you’ll see options to log out or terminate that specific session. Unlike ChatGPT, you can remove individual sessions without affecting your other logins. There is no global log-out-all button, so you’ll need to review and terminate each suspicious session individually.

Why Claude Doesn’t Need Password Resets

Because Claude doesn’t use passwords, there’s nothing to change after a breach. Once you’ve terminated all unauthorized sessions, your account is secure again. You regain access simply by logging back in with your email address, which triggers Claude to send you a new login link. This design eliminates the risk of password reuse, password weakness, or credential stuffing attacks that plague traditional authentication systems.

Securing Your Perplexity Account

Perplexity’s approach is more limited in granularity but remains effective for emergency situations.

The Limitation: No Session Visibility

Perplexity does not provide a dashboard showing where your account is currently logged in. This means you cannot see individual sessions or verify which devices have access. However, it also means you won’t spend time trying to identify which of many sessions might be suspicious—if you’re concerned about a breach, you have one clear action available.

The Nuclear Option: Sign Out Everywhere

Open Perplexity in your browser and click your username in the bottom-left corner. Navigate to All settings, then locate and click Sign out of all sessions. A confirmation prompt will appear—click Confirm to log out from every device simultaneously. This action terminates every active session instantly, forcing both you and any attacker to authenticate again.

Logging Back In After a Full Logout

Once you’ve signed out globally, visit Perplexity and enter your email address to begin the login process. You’ll receive an email containing a unique six-digit code. Enter this code on the Perplexity website to log in, or click the Sign in button in the email itself for direct access. Because Perplexity uses email-based authentication similar to Claude, there’s no password to reset—regaining access is as simple as verifying your email.

Universal Best Practices for AI Platform Security

Across all three platforms, several preventive measures dramatically reduce your risk of being hacked in the first place. Use a unique password for each AI service you access, never reusing credentials from other accounts or services. A password manager like Bitwarden, 1Password, or Dashlane makes managing unique passwords painless. Enable multi-factor authentication wherever it’s available—ChatGPT and Perplexity both support it, adding a second verification step that protects you even if your password is stolen. While Claude currently doesn’t offer MFA, the email-link authentication provides inherent protection against password attacks.

Beyond these account-level protections, stay vigilant about email security. Since all three platforms rely on email for authentication or password recovery, an attacker who compromises your email account gains leverage over your AI accounts as well. Use a strong, unique password for your email account, enable MFA there, and monitor your email login activity regularly. If you use Gmail, Google provides a security checkup tool; Microsoft Outlook users can check their account security and recent activity similarly.

Review your session activity regularly, even if you haven’t noticed anything suspicious. Setting a monthly reminder to check your active sessions on ChatGPT and Claude takes minutes but catches intrusions early. For Perplexity, since you can’t see ongoing sessions, consider doing a full sign-out every few months as a precautionary measure, or whenever you access the service from an unfamiliar network like a coffee shop or hotel WiFi.