Hackers used Meta AI chatbot to hijack high-profile Instagram accounts including Obama White House

Hackers Bypassed Meta AI to Hijack High-Profile Instagram Accounts

Hackers exploited Meta's AI chatbot to gain access to high-profile Instagram accounts—including the Obama White House account and a Space Force account—by simply asking the chatbot to send password reset codes to attacker-controlled email addresses. The chatbot complied without proper validation.

The exploit reveals a fundamental security flaw in offloading account recovery to AI systems without segmentation or verification controls. When someone asked Meta AI for access to a specific account and requested a reset code be sent to an attacker's email, the system obliged. There was no check to confirm the requester owned the account or had legitimate access rights.

The targeting was selective, focusing on one-word Instagram usernames that typically belong to high-profile accounts. The attack surface was broad enough to compromise official government accounts, suggesting the vulnerability affected a significant class of users before remediation.

The incident underscores a broader risk as platforms increasingly route sensitive support functions through AI systems: without proper authentication, rate-limiting, and access controls, AI chatbots become a direct vector for account takeover rather than a tool to prevent it.