Executive Office of the President

Summary

Hackers exploited a vulnerability in Meta’s AI‑powered Instagram support chatbot to reset passwords and seize control of accounts, affecting roughly twenty thousand users. The flaw allowed attackers to link their own email to a target account via the chatbot, receive a verification code, and change the password when two‑factor authentication was not enabled. Among the compromised profiles was the former White House Instagram page, along with numerous other high‑profile accounts such as a major retailer and a senior military official. Meta disabled the chatbot, patched the verification check, forced password resets on affected accounts, and placed them in a mandatory security checkpoint while preparing user notifications.

Motives

Detailed motive labels are available to members.

TTPs

Detailed technique labels are available to members.

Description

On May 31, 2026, Meta discovered that its High Touch Support (HTS) tool, an AI‑powered account recovery chatbot, was being exploited. Attackers used the chatbot to link a victim’s Instagram account to an email address they controlled. After providing a verification code sent to that email, the chatbot displayed a password‑reset option. To increase success, hackers first connected to a VPN server near the victim’s suspected location. The tool then sent the verification code to the attacker’s inbox, allowing them to reset the password. If the account had no two‑factor authentication enabled, the attacker could log in after the reset. Among the compromised profiles were the dormant Obama White House Instagram account (inactive since 2017), the Sephora brand account, the US Space Force Chief Master Sergeant John Bentivegna account, and security researcher Jane Wong’s personal account. The exploit was shared in videos and instructions on platforms such as X and Reddit. Meta later disclosed to the Maine Attorney General’s Office on June 8 that 20,225 individuals were potentially affected. Meta noted that the actual number could be smaller because some password‑reset events may have been performed by legitimate account holders.

Meta stated that it is unclear whether personal data stored in the compromised accounts was actually accessed. However, the company warned that attackers could have obtained profile information, email addresses, phone numbers, dates of birth, direct messages, social‑media posts, and interaction histories. The takeover of high‑profile accounts led to some of them being offered for sale on dark‑web marketplaces. In response, Meta forced all affected users to reset their passwords to regain control. The company also placed the affected accounts in a mandatory security checkpoint. All password‑reset links generated by exploiting the vulnerability were invalidated. The abused High Touch Support tool was taken offline and will remain disabled until the underlying bug is fixed. Meta said it will repair the authentication check in the Instagram recovery entry point so that any future request to link an email address is verified against the account’s existing information before a password reset is allowed. Once the fix is verified, Meta plans to re‑enable the support chatbot. Meta intends to send user notifications to the potentially impacted individuals about the incident.

Meta’s associate general counsel Amber Hannah told the Maine Attorney General that the chatbot functioned as designed but a bug in a separate code path caused it to fail to verify that the supplied email matched the account’s email. The company’s spokesperson Andy Stone confirmed on X that the vulnerability had been patched and the issue resolved. Prior to the incident, Meta had launched its AI support assistant in March 2026, describing it as a 24/7 help tool for password and profile‑setting issues. In the weeks before the hack, Meta had laid off approximately 8,000 employees, including members of its integrity and cybersecurity teams. After the breach, Meta’s internal review concluded that the exploit stemmed from insufficient validation of email ownership during the password‑reset flow. The company has not disclosed any evidence that the attackers accessed the contents of direct messages or other private data beyond what is possible through a password reset. Meta continues to monitor the affected accounts for any signs of further unauthorized access. No further details about the attackers’ identities or motives have been released by Meta or law‑enforcement agencies. Meta has committed to implementing stronger verification mechanisms before restoring the chatbot to full service. The support chatbot will remain offline until those mechanisms are in place and tested.

Sources

Sources available to members: 4 sources.