Attackers used Meta’s Instagram AI support bot to reset passwords and hijack accounts

Attackers used Meta’s automated Instagram support assistant to take over accounts, including the Obama White House account and the U.S. Space Force chief master sergeant account, and briefly deface them with pro-Iran messages. According to KrebsOnSecurity and Telegram posts cited in the report, the abuse involved the password-recovery flow: attackers asked the AI bot to add a new email address to a target account, then used the one-time code sent there to reset the password. No CVE is given, Meta reportedly pushed an emergency patch, and accounts with multi-factor authentication enabled were said to resist the takeover.
Why it matters: This matters because it shows AI-driven customer support can become a new social-engineering path to account takeover even without a backend database breach. Instagram users, especially high-value or public-facing accounts, should enable multi-factor authentication now and review account recovery email addresses and recent login activity.

Sources

Meta Says 20,000 Instagram Accounts Hacked via AI Tool Abuse
Eduard Kovacs 2026.06.08 98% relevant
This is the same underlying event: abuse of Meta’s AI-powered Instagram account recovery/support workflow to reset passwords and hijack accounts. It adds Meta’s disclosure that 20,225 accounts were potentially affected, the discovery date (May 31), a precise explanation of the email-verification bug in the High Touch Support tool, and remediation steps including disabling the tool, invalidating reset links, and forcing security checkpoints.
Over 20,000 Instagram accounts stolen in Meta AI support hack
Sergiu Gatlan 2026.06.08 99% relevant
This is the same underlying event: abuse of Meta's High Touch Support AI-assisted Instagram recovery flow to issue password reset links and hijack accounts. The article adds Meta's breach disclosure, an estimated impact of over 20,000 stolen accounts, timeline details including discovery on May 31 and breach activity dating to April 17, and Meta's response steps such as disabling HTS, invalidating reset links, and requiring account re-authentication.
Hacking Meta’s AI Chatbot
Bruce Schneier 2026.06.04 98% relevant
This is the same underlying event: attackers abused Meta’s Instagram AI support assistant to add attacker-controlled email addresses, receive verification codes, and trigger password resets for victim accounts; this source adds that Meta spokesperson Andy Stone said the issue was fixed.
Instagram users locked out after Meta AI abused to steal accounts
Bill Toulas 2026.06.02 99% relevant
This is the same underlying event: attackers abused Meta’s AI-powered Instagram support and recovery process to change account email addresses, bypass recovery safeguards including selfie verification and reportedly 2FA, and hijack high-value accounts such as the Obama White House account. This source adds reporting on victims being trapped in AI-only recovery loops, claims that AI-generated animated selfies were accepted for identity checks, and Meta communications VP Andy Stone’s statement that the issue was resolved and impacted accounts were being secured.
Meta AI Hands Over High-Profile Instagram Accounts to Hackers
Ionut Arghire 2026.06.02 99% relevant
This article covers the same underlying event and adds specifics on the attack path: a confused-deputy logic flaw in Meta’s AI-powered recovery assistant let attackers relink victim accounts to new email addresses, use VPNs to mimic victims’ locations, sometimes submit AI-modified selfies, and then reset passwords without effective 2FA blocking. It also says Meta has now fixed the issue.
Hackers Used Meta’s AI Support Bot to Seize Instagram Accounts
BrianKrebs 2026.06.01 100% relevant
This article appears to be the first concrete report tying a specific Meta AI support-bot recovery flaw to real Instagram account hijackings and visible defacements.
← Back to all stories