
Send Email
Confidentiality Guaranteed
Confidentiality Guaranteed

Identity theft
Hackers hijacked high-profile Instagram accounts by asking Meta’s AI crimson meat up chatbot to alternate story electronic mail addresses with out identity verification. Meta says the flaw is mounted, but attacks reportedly continued after the corporate’s announcement.
No phishing link. No malware. No SIM swap. Hackers took over high-profile Instagram accounts over the weekend by doing something disarmingly easy: they asked Meta’s AI buyer crimson meat up chatbot to alternate the electronic mail tackle on anyone else’s story. The bot complied with out verifying the requester’s identity, and the attacker then reset the password and locked out the rightful proprietor.
The system, which became as soon as first reported by 404 Media, unfold through Telegram channels the build hackers shared the procedure in which and started promoting stolen handles for sale. Among the many compromised accounts had been the dormant Obama White Residence Instagram profile, which became as soon as old-fashioned to post unauthorised AI-generated photos, and the story of US Space Force chief master sergeant John Bentivegna.
Meta spokesperson Andy Stone acknowledged on Monday that “the say that did happen has already been mounted.” But on Tuesday, more Instagram users reported shedding gain entry to to their accounts, and participants of the same Telegram channels claimed the exploit aloof labored, essentially based totally on TechCrunch.
The system exploited a flaw in Meta’s AI Support Assistant, which the corporate rolled out in March 2026 with the ability to “unravel story disorders from launch to withhold out,” including resetting passwords. The chatbot became as soon as designed to exchange human crimson meat up brokers for routine story restoration duties.

The 💜 of EU tech
The most new rumblings from the EU tech scene, a legend from our smart ol’ founder Boris, and a few questionable AI art. It is free, every week, in your inbox. Ticket in now!
An attacker would establish a target story, on the total a brief “OG” username price hundreds on underground markets. They would per chance use a VPN to spoof the target’s presumed location, launch a chat with the AI crimson meat up bot, and merely claim to be the story proprietor. The bot would then link the attacker’s electronic mail tackle to the target story with out asking for any proof of ownership.
A human crimson meat up agent would beget verified the caller’s identity earlier than making this kind of alternate. The chatbot didn’t. Two-say authentication would per chance maybe beget blocked some takeovers, but accounts with out it enabled had been at chance of compromise in minutes.
For years, a flourishing underground market has existed for thus-called OG usernames, the immediate, neat handles claimed by Instagram’s earliest users. Outdated systems of stealing them required technical sophistication: phishing the victim, bribing telecom insiders to plot SIM swaps, or compromising electronic mail accounts.
This attack diminished the barrier to entry dramatically. The hackers who shared the system on Telegram had been promoting it sounds as if stolen handles for sale, including linked old forenames and nation names that feature as collectibles in this grey market. TechCrunch reported that the gross sales continued even after Meta’s announced repair.
Meta has been sending password reset emails and security notifications to users whose accounts had been focused. Plenty of victims reported receiving messages from Instagram warning that the corporate had “detected some suspicious exercise that means your Instagram would per chance maybe had been compromised,” along with instructions to reset their passwords.
Stone told TechCrunch that Meta secured affected accounts on Monday earlier than origin its notification campaign. He declined to articulate what number of users had been compromised. Meta furthermore disputed that the Obama White Residence story became as soon as taken over the use of this particular procedure, though it confirmed the story became as soon as hacked.
The incident exposes a fundamental tension in deploying AI brokers with valid-world authority. Meta constructed its crimson meat up chatbot to plot actions that beforehand required a human within the loop, but it shipped that functionality with out the verification checks that human brokers would beget applied as a topic for sure.
It is a pattern the industry has seen earlier than. When Instagram story restoration became as soon as handled by humans, the system became as soon as unhurried and veritably frustrating, but it now now not lower than required the requester to expose they had been who they claimed to be. Automating that process with out conserving the identity-verification step grew to change into a bottleneck valid into a vulnerability.
The broader lesson is now now not that AI must never take care of sensitive story operations, but that authentication stays a topic topic no chatbot can shortcut. Meta gave its AI the vitality at quit the keys. The hackers merely walked up and asked for them.
