Sev0

AI security incidents, tracked

All incidents / Leaky AI product

Meta's AI support tool abused to hijack 20,000 Instagram accounts

Meta (Instagram), disclosed Jun 2026 Leaky AI product Data exposed Medium

What happened

Attackers abused Meta's AI-assisted account recovery tool, which failed to check that the reset email matched the account, to take over 20,225 Instagram accounts without two-factor authentication. Meta disclosed the incident in a breach notice and says affected accounts were secured.

High Touch Support is an AI-assisted tool that helps Instagram users who are locked out regain access, including by sending a password reset link to an email address. A bug in a separate code path meant it never checked that the email address supplied belonged to the account, so it would send a valid reset link to an address chosen by the requester. According to reporting by 404 Media, attackers shared the method on Telegram: they connected through a VPN to appear to be in the same country as the target, started a reset and asked the AI assistant to link their own email to the account. Accounts without two-factor authentication could then be taken over, exposing their messages, contact details and profile data. The abuse ran from about 17 April until Meta found it on 31 May, and reported victims included the Obama White House archive account, a senior US Space Force official and Sephora.

How it happened

Leaky AI product: App flaw. A security flaw in an AI application, such as broken login or access checks, let someone reach data they should not have seen.

The AI involved was from Meta. The product was Meta AI support assistant, a type of AI-powered app.

Impact

Data was exposed. Data involved: personal data. The impact was confined to one organization and its users or customers. Sources give a figure of 20,225.

Severity is rated Medium, calculated from how far the incident reached and how sensitive the data was.

Was the AI at fault?

No. The failure was in the systems, settings or people around the AI, not in the model's behaviour.

Status and timeline

Confirmed by the affected organization, the AI vendor, a regulator or a named security research firm.

It happened or began on 17 Apr 2026. It became public in Jun 2026.

Meta disabled the vulnerable functionality and invalidated outstanding reset links on the day it found the problem, secured the affected accounts, notified regulators including Maine's attorney general, and said it was reviewing similar recovery processes across its platforms.

Sources

  1. https://www.bleepingcomputer.com/news/security/meta-ai-support-data-breach-affects-20-000-instagram-accounts/

Related incidents

Cite as: Sev0, "Meta's AI support tool abused to hijack 20,000 Instagram accounts", https://sev0.fyi/incidents/2026-06-01-meta-ai-support-instagram-takeovers/. Data for this record: incidents.json.