Meta $META +0.79%'s AI support chatbot was exploited by hackers to take over a wave of Instagram accounts, including those belonging to high-profile users, before the company patched the vulnerability, according to TechCrunch.
Crucially, the method worked without the attacker ever controlling or accessing the email address the victim had on file with Instagram. To execute the takeover, an attacker would initiate a chat with Meta's AI Support Assistant and instruct it to register a different email address on the victim's account. A verification code would arrive at the address the attacker had provided; relaying that code back to the bot was enough to unlock a password reset prompt. Rounding out the technique was a VPN connection set to the victim's geographic region — since Meta's support system used physical location as an authorization signal, spoofing that detail was sufficient to keep Instagram's automated defenses from intervening, according to Engadget.
Among the compromised accounts were an Obama-era White House handle that had been inactive since 2017, the Instagram account of U.S. Space Force Chief Master Sergeant John Bentivegna, beauty retailer Sephora, and security researcher Jane Wong. "The password got changed without my knowledge and I was getting different password reset attempts throughout yesterday," Wong told TechCrunch. "Quite concerning."
