Hackers successfully tricked Meta's AI-powered support bot into handing over control of several high-profile Instagram accounts, according to a report from 9to5Mac. Among the compromised accounts were those belonging to the White House, the US Space Force, and security researcher Jane Wong.
The attack targeted Meta's automated support system, exploiting vulnerabilities in how the AI handles account recovery requests. This incident raises serious questions about relying on artificial intelligence for security-sensitive tasks, particularly on platforms with billions of users.
Only one source, 9to5Mac, has reported on this breach. The specific method used by the hackers has not been disclosed, and Meta has not yet commented on the incident or whether it has patched the vulnerability.
Users of Instagram, especially those with large followings, may face increased risk until Meta addresses the flaw. The breach also underscores broader concerns about AI systems being manipulated in high-stakes environments.
Some experts argue that AI bots can improve response times and reduce human error in support tasks, but this incident shows they also introduce new attack surfaces that traditional systems may not have.