The threat involves exploitation of a confused deputy vulnerability in Meta's AI chatbot used for Instagram account recovery. The AI assistant had API access to account management functions and was tricked into linking targeted accounts to new email addresses controlled by attackers. This enabled password resets and account takeovers without proper verification, bypassing 2FA and fraud detection mechanisms. Attackers used VPNs to spoof geographic locations and AI tools to manipulate photos for selfie verification. Hundreds of high-profile accounts were compromised before Meta fixed the flaw.

Attackers gained unauthorized access to high-profile Instagram accounts, including notable public figures and organizations, by hijacking account recovery processes. This resulted in account takeovers, locking out legitimate owners, and subsequent sale of compromised accounts on the dark web. The exploit bypassed two-factor authentication and fraud detection, increasing the severity of the impact. The exact number of affected accounts remains unknown.

Meta has officially resolved the vulnerability in their AI assistant, and the exploit no longer works. Organizations and users should ensure their accounts are secured with updated protections and monitor for any suspicious activity. Since the issue is fixed, no immediate action is required beyond standard security vigilance. Check Meta's official communications for any further guidance.