Overview
- Security researchers and videos circulated on Telegram showed attackers used Meta’s AI support assistant to add attacker‑controlled emails to Instagram profiles and obtain verification codes for password resets.
- The exploit, which spread over the weekend, enabled brief takeovers of several high‑profile accounts including the Obama‑era White House archive, Sephora, and a U.S. Space Force chief master sergeant.
- Meta’s communications lead Andy Stone said the company pushed an emergency patch and is securing impacted accounts, but Meta has not disclosed how many users were affected.
- Reports indicate attackers used VPNs or residential proxies to spoof location and repeated the automated flow until it succeeded, and accounts with multi‑factor authentication were generally protected.
- Researchers warn the incident exposes risks of giving AI direct authority over sensitive recovery workflows, and how‑to materials are circulating on Telegram and black‑market channels, prompting calls for stronger verification and audits.