Meta's AI Chatbot Bug Exposed Over 20,000 Instagram Accounts to Hijacking

Summary

Meta confirmed that over 20,000 Instagram accounts were compromised through a critical vulnerability in its AI support chatbot between May 31 and June 1, 2026. The bug allowed attackers to bypass standard account verification by requesting password resets to email addresses not associated with the target accounts, which the system incorrectly accepted instead of rejecting. High-profile victims included former President Barack Obama's White House account, a US Space Force official, and the Sephora brand account.

The vulnerability stemmed from a faulty code path in Meta's AI chatbot that failed to properly verify whether an email address provided during a password reset request matched the account owner's registered email. Rather than rejecting mismatched addresses, the system sent password reset links to unverified email addresses, enabling unauthorized third parties to gain account access without two-factor authentication. Meta responded by disabling the AI support tool entirely, removing the buggy code path, invalidating malicious password reset links, and enrolling all potentially affected accounts in mandatory security checkpoints.

While Meta stated it is unaware of any confirmed personal data access, compromised accounts could have exposed sensitive information including email addresses, phone numbers, birthdates, social media posts, direct messages, profile information, and connected account details. The company filed a formal breach notification notice with the state of Maine, identifying 30 affected residents in that state, though Meta acknowledged this figure may be an upper estimate of actual incidents.

• Meta disabled the compromised tool, removed the buggy code, and enrolled affected accounts in mandatory security verification
• Potentially exposed user data includes emails, phone numbers, birthdates, DMs, posts, profile information, and connected account details