Meta’s AI-powered support chatbot was exploited by hackers to take over Instagram accounts, with the vulnerability allowing attackers to change account email addresses and lock out original owners. Meta says the issue has since been patched.
The exploit was documented in a video shared on Telegram, in which a hacker asks Meta’s support chatbot to link a new email address to someone else’s account. The chatbot then sends a verification code to the hacker’s email, which they use to set a new password — effectively seizing control of the account. Some attackers used a VPN to spoof their location to match that of their target while contacting Meta support.
Meta rolled out its AI support assistant in March 2026, designed to help users reset passwords, set up two-factor authentication, and regain account access. The vulnerability appears to have been exploited around the same time Barack Obama’s @obamawhitehouse Instagram account was hijacked and used to post Iranian propaganda. Hackers also appeared to have taken over accounts belonging to the US Space Force Chief Master Sergeant and beauty retailer Sephora, according to 404 Media.
Attackers appeared to target high-value usernames — short handles consisting of a single letter or word, such as “h” or “eggs.” Security researcher and reverse engineer Jane Manchun Wong said her own account was compromised, writing on X that her password was changed without her knowledge and that she was repeatedly logged out of the Instagram iOS app.
Gergely Orosz, creator of The Pragmatic Engineer newsletter, wrote on X that Instagram’s trust and safety team had been “absolutely gutted” in recent weeks due to layoffs and reassignments to AI labeling tasks. “Apparently this was not a sophisticated hack,” Orosz wrote, adding that engineers had “no incentives for stuff like… security.”
The incident suggests that deploying AI tools in account security workflows may carry significant risks if those systems can be manipulated into bypassing standard verification processes.
Source: The Verge