What Happened: An AI Support Bot Turned Account Takeover Tool
The Meta AI security flaw was an account takeover vulnerability in Instagram’s AI-powered support system that allowed attackers to reset passwords and hijack accounts using text prompts, without needing the victim’s email, phone, or existing password. Reports surfaced across Reddit and X from people who saw their Instagram account hacked, logged out repeatedly, or bombarded with password reset attempts. Security researcher Jane Manchun Wong said “the password got changed without my knowledge,” highlighting how silent and sudden the hijack could be. High‑value and short “OG” usernames were popular targets, along with prominent profiles such as the inactive Obama-era White House handle and the account of U.S. Space Force Chief Master Sergeant John Bentivegna. Behind the scenes, attackers abused Meta’s AI chatbot as if it were an overly helpful support agent that failed to confirm identity before changing critical account settings.
How the Password Reset Exploit Worked Step by Step
The password reset exploit relied on persuading Meta’s AI Support Assistant to change the email linked to a victim’s Instagram account. Attackers typically started by turning on a VPN to appear near the target’s location, making Instagram’s automated risk systems less likely to flag their access. Next, they opened a chat with the AI chatbot and claimed to be the account owner, asking it to attach a new email address they controlled. The bot accepted this at face value and sent a verification code to the attacker’s inbox. After the hacker sent that code back to the bot, the AI presented a Reset Password option, allowing a full account takeover without the victim’s credentials. In many cases, this flow bypassed two-factor authentication because the attacker had already switched the account’s primary email to their own address.
Why the First Fix Failed: A Button Gone, a Backend Still Open
Meta announced that the issue with its AI support assistant had been fixed, but users kept reporting their Instagram account hacked even after that statement. Developers and security watchers say the company removed a visible “Get Support” button in the interface, but left the underlying API endpoints exposed. This meant attackers could still send crafted text prompts to Meta AI and reach the vulnerable flows that changed email addresses and enabled password resets. Instagram’s Trust and Safety team had reportedly been cut back heavily as staff were shifted toward AI projects, leaving fewer people to review risky automation. As a result, the chatbot behaved like an inexperienced employee: eager to help, too trusting of what it was told, and unaware of how attackers trade account takeover techniques in Telegram channels and other gray markets.
Who Was Hit and How Meta Is Responding Now
The campaign went after a wide mix of accounts, from long‑abandoned profiles to high‑profile names and collectible short handles. TechCrunch and other outlets reported that even the dormant Obama-era White House account and the profile of Space Force Chief Master Sergeant John Bentivegna were taken over. Many victims described multiple password reset emails and forced logouts before losing access entirely. After initial claims that the flaw was resolved, more account takeover reports appeared the next day. Meta spokesperson Andy Stone later said the company had “already secured impacted accounts,” and added that some people would receive password reset notifications while others might see security questions during login. Meta has begun emailing users it believes were affected and locking down compromised profiles. The company also disputes some specific targets, but has not shared how many accounts were hijacked overall.
If Your Instagram Account Was Hacked: What to Do Now
If you suspect your Instagram account hacked in this incident, start by using Instagram’s official account recovery tools, not any AI chatbot. Check your email for alerts from Meta about suspicious logins, email changes, or password resets, and act on any recovery links quickly. Once back in, confirm that the email address and phone number on your account are yours and up to date, and turn on two‑factor authentication using an authenticator app rather than SMS where possible. Review active sessions and log out devices you do not recognize. Avoid searching for help through random “support” chats, Telegram groups, or third‑party services that offer to recover OG handles, as these often lead to more account takeover vulnerability risks. Finally, be cautious with the Meta AI assistant: treat it as a helpful but fallible tool, not as a guaranteed secure gatekeeper for your password reset exploit defense.
