Meta says hackers will no longer be able to access accounts with its AI chatbot – National
Meta says it has resolved an issue that allowed hackers to trick its AI assistant into giving access to the accounts of other users, including high-profile people, according to multiple media reports.
404 Media and The Guardian say hackers used the trick to target Barack Obama's White House account, beauty retailer Sephora and US Space Force master sergeant John Bentivegna.
Regular users also reported similar exploits on X, even sharing video recordings of conversations with Meta's AI chatbot explaining how the hacks were executed.
One video posted on X appears to show a hacker asking the Meta AI assistant to link his account to a new email address, bypassing two-factor authentication.
Get the latest country news
Get the best Canadian news delivered to your inbox as it happens so you never miss a trending story.
The bot replies that a verification code has been sent to the new email address and asks the hacker to enter it in the chat. Once the hacker has done so, an option appears that allows them to reset the stolen user's password.
“This issue has been resolved and we are recovering the affected accounts,” Meta said in a statement to Global News on Tuesday.
The breach comes as Meta changes operations that will see human roles transferred to AI and the increased use of tech-forward features across its platforms, including customer support on Instagram and Facebook.
The company's chatbot customer support feature, introduced last year, will continue to be updated over time, according to a March update from the tech giant.
Meta said in its release that the AI chatbot is designed for users to report scams, fake accounts or problematic content as well as manage privacy, reset passwords, and update profile settings.
Last year, Meta reported “positive results” from changes to the service, which it said reduced errors and focused on controlling illegal and harmful content across Meta's platforms, including terrorism, child exploitation, drugs, fraud and scams.
Meta CEO Mark Zuckerberg arrives for a landmark lawsuit over whether social media is intentionally addictive and harmful to children, Feb. 18, 2026, in Los Angeles.
AP Photo/Ryan Sun
AI systems will “eventually be able to do better technical work, such as iterative review of graphic content or areas where adversary actors are constantly changing their strategies, such as illegal drug sales or scams,” noted the March release, before adding that Meta “will still have people reviewing content.”
In January, hundreds of Meta employees were laid off from its Reality Labs division as the tech giant shifted its focus from metaverse products to AI. In late May, Reuters reported that CEO Mark Zuckerberg said in a company-wide memo that the company was not planning any more layoffs in 2026.
He made the announcement on the same day that Facebook's owner made a major restructuring of the company, laying off 10 percent of its workforce worldwide and transferring 7,000 more employees to new programs related to AI workflows.
– With files from Reuters
© 2026 Global News, a division of Corus Entertainment Inc.
