AI chatbot of Meta has helped attackers take Instagram accounts

Meta's AI chatbot could be used by third parties with simple requests to control any account on Instagram. Details have now been announced after, among other things, an old account of the White House has shared Iranian propaganda under Barack Obama. Whether there is a connection, however, is unknown. The attack route was made public by the US-Techmagazin 404 Media, which is based on Telegram groups of security researchers and hacking groups. The Instagram Group Meta has already reacted, the Group spokesperson assured X: “The problem has been solved and we secure affected accounts.”

Scary simple method

According to the report, you only had to pretend by VPN that you are in the home region of the person whose Instagram account is to be captured. Next, an AI chatbot for support queries could be asked to connect the desired account with a new email and then send a code to reset the password there. The chatbot had followed him, reports 404 Media. How often such attacks have been successfully carried out is unclear. On the short message service X, however, the well-known app researcher Jane Manchun Wong explained that this method changed the password of her account.

Video by heise

Meta released an AI assistant for support requests only in March, which should help with problems on Facebook and Instagram around the clock. At the time, the Group stated that the AI should help “update the password and change the settings in the profile”. The AI assistant was designed to solve problems with an account and not just make suggestions, it said at the time. 404 now writes that long lists with account names were placed on Telegram, which were allegedly taken over by using this function. It can't be checked. The attack method shows the enormous risk associated with the outsourcing of technical support to AI technology, writes the portal.

(mho)