A Meta‑created AI bot hacked Instagram accounts, the service told users.

Cyberattack via Meta AI chat‑bot: how attackers “stole” Instagram accounts

Over the past weekend cybercriminals revealed a method for taking control of popular Instagram pages using Meta AI’s official support chatbot. Victims were account owners with short unique usernames (e.g., @coolcat) who complained about losing access.

How the scheme works
1. Contacting the bot

The hacker tells the chatbot that they own the target account and asks it to “link” the account to their e‑mail.

2. Bot processes the request

The AI, without extra checks, fulfills the request, changing the linked email to the attacker’s address.

3. Password reset

After the email is changed, the hacker resets the password and gains full control of the page. The account owner loses access and their e‑mail is blocked.

No Meta employees or contractors were involved – everything happened automatically through AI.

Company reaction
- June 1: Meta spokesperson Andy Stone announced that “the issue has already been resolved.”

- June 2: the number of complaints increased. Stone clarified that some users receive password‑reset notifications, while others encounter security questions when trying to log in.

Meta began protecting affected accounts immediately after the announcement, but the exact number of compromised profiles was not disclosed.

Why this became possible
In March Meta launched an AI chatbot in its support section and granted it a portion of powers previously reserved for humans. Among them was the ability to “safely reset passwords.”

The market for short Instagram usernames has existed for several years: sellers offer them as collectible items. Previously hacking such accounts required phishing, phone number hijacking, or even bribing insiders. Now it’s enough to use a “cool” AI chatbot.

Key facts

- Hackers use Meta AI chat‑bot to change the email linked to an account.
- After resetting the password they gain full control of the page.
- Meta announced a fix on June 1, but complaints intensified the next day.
- The AI was granted powers that had previously belonged only to humans, allowing attackers to bypass traditional security measures.