Meta confirmed that thousands of Instagram accounts were hacked by exploiting AI chatbots.

Meta is alerting thousands of people whose Instagram accounts have been compromised following months of abuse of the company's AI chatbot, which hackers repeatedly tricked into taking control of individuals' accounts.

In a new data breach notification letter seen in this week's Security, Meta revealed for the first time how many people had their accounts compromised as part of a long-running hacking campaign discovered earlier this week and first reported by 404 Media ($) and TechCrunch ($). The number of accounts affected provides some clarity on how widespread this hacking campaign is and how long it has been running.

According to a data breach notification filed with the Maine Attorney General's Office late Friday, Meta notified at least 20,225 people, including 30 in Maine, that their accounts had been compromised.

This compromise allowed the hackers to take over the person's entire Instagram and all associated accounts, including obtaining contact information, date of birth, and profile information, as well as access to the person's posts, direct messages, and account activity.

According to Meta's notice, the breach was confirmed to be related to a "vulnerability in the AI-assisted account recovery system for Instagram" that was exploited to "perform password resets on Instagram user accounts."

As previously reported, hackers exploited a flaw in the Meta chatbot that allowed anyone to reset passwords for accounts that didn't have two-factor authentication turned on. The bug tricked the chatbot into sending a verification code to an email address controlled by the hacker rather than the account holder's email address on file. Anyway, the chatbot responded.

"The tool itself was functional and performed as intended. However, a bug in a separate code path prevented the system from properly verifying that the email address provided by an individual requesting a password reset matched the email address associated with that user's Instagram account," Meta said in its breach notice.

"As a result, when individuals provided an email address that was not previously associated with an account, the system incorrectly sent a password reset link to the unconnected email instead of rejecting the request. This allowed unauthorized third parties to receive password reset links for accounts they did not own," the company added.

At this point, Meta says, a hacker can reset someone's password and take over their account as if they were the rightful owner.

Mehta said he had "no idea" what personal information was accessed during the hacking process. (An email sent to Meta's press line requesting clarity on this had not been returned as of early Saturday.)

According to Maine's listing, the hack began around April 17 and continued until this week, when Meta announced it had secured the chatbot. Instagram reportedly began notifying affected individuals by sending out password reset notifications earlier this week, despite some reports that a hack was in progress.

Meta also confirmed in its notice that it had "instructed affected users to reset their passwords and re-authenticate through secure and verified channels" and warned users to protect their accounts.

Meta said it has now disabled its AI chatbot, removed code paths that allow the chatbot to reset user accounts, and is also checking other chatbots across the platform to prevent repeat incidents. It is not yet clear under what circumstances chatbots have been abused. But it comes shortly after Meta laid off thousands of employees and rewarded top executives with stock incentives as the company continued to double down on AI.

~This week’s security episode~ Thank you so much for reading. If you liked this article, please share it! If you have any feedback, questions or comments about this article, please feel free to contact us at this@weekinsecurity.com.