First Hint Something Was Awry: New IG Posts From 'Obama'

Instagram's long-dormant White House account for former President Obama recently came roaring back to life, complete with bizarre, out-of-the-norm posts about President Trump and ones that insisted the White House was "under Shiite control." Per the New York Times, that wasn't Obama making those posts, but a group of hackers that found a flaw in a Meta customer-service tool that allowed them to use an AI chatbot to reset passwords on Instagram accounts simply by asking it to do so. About 34,000 accounts were targeted and some 20,000 were breached, including those belonging to a senior Space Force official and the beauty brand Sephora, per 404 Media, which originally reported on the hack.

In several cases, hijacked accounts pushed pro-Iran messages and other political content, per the Times. Meta's internal documents say usernames were changed on more than 3,500 accounts, and personal data like email addresses, phone numbers, and birthdates may have been exposed, though the company says it can't tell exactly what was accessed. Meta says it has since patched the issue, secured impacted accounts, and notified regulators and users.

A spokesman blamed failed "back-end checks," not the AI agent itself, and pointed to a 30% rise in successful account recoveries since rolling out automated tools. Despite the incident, Meta is pressing ahead with its AI-heavy strategy, including new chatbots for businesses across Instagram, WhatsApp, and Messenger. Cybersecurity pro Ian Goldin says we likely haven't seen the last of this. "AI chatbots create interesting new attack [surfaces]," he tells Krebs on Security. "We're likely going to see a lot more of these kinds of attacks."