Meta Confirms AI Bug Exposes Instagram Accounts

Meta confirmed that a vulnerability in an AI-powered account-recovery tool allowed unauthorized access to Instagram accounts, with the company notifying at least 20,225 affected users, according to Infosecurity Magazine, BleepingComputer, Help Net Security, and Security Affairs. Reporting identifies the tool as an AI-assisted "High Touch Support" (HTS) system. Per that coverage, a separate code path failed to verify that the email address provided during a password reset matched the address on the target account, so the system issued reset links to email addresses that were not associated with the account.

According to the reporting, an attacker could request a reset for an account they did not own, receive the reset link at an attacker-controlled email, and complete the takeover when the rightful owner had not enabled two-factor authentication. Coverage says the flaw was active from approximately April 17, 2026 until Meta disabled the tool in early June, and that attackers focused on high-profile accounts (including, per reports, the Obama White House account and a U.S. Space Force account) and short, high-value usernames that can be resold.

Per the reporting, Meta disabled the AI-assisted HTS tool and the vulnerable code path, invalidated all outstanding password-reset links, and placed affected accounts into a mandatory security checkpoint that blocks access until additional verification is completed.

account-recovery and password-reset flows are repeatedly targeted because they provide a path that can bypass primary authentication. Adding AI-assisted automation or decision logic to those flows can enlarge the attack surface by introducing new code paths and dependencies, and this incident illustrates how a verification gap in a support tool can scale to tens of thousands of accounts. Two-factor authentication materially limited impact here, reinforcing its role as a backstop when recovery logic fails.

Editorial analysis: useful signals include any detailed post-mortem from Meta on the specific component and root cause, follow-up advisories or indicators of compromise, and renewed guidance on secure account-recovery design. The episode is also likely to prompt scrutiny of AI-assisted support tooling and its verification controls across consumer platforms.