More and more companies are replacing customer service representatives with artificial intelligence (AI) chatbots because they can work non-stop and respond to messages quickly regardless of time. But Meta’s use of chatbots led to user accounts being easily stolen over the weekend.
Among the victims were the accounts of Barack Obama’s White House administration, Sephora and the head of the US Space Force. Hackers took advantage of the vulnerability in Meta AI’s chatbot. They simply asked for Meta AI’s help because they were the real owners of the accounts they wanted to steal.
The hackers asked for the account’s email to be transferred to their own email and Meta AI only asked for a selfie video to be taken to verify identity. The hackers then used generative AI to generate a selfie with the account owner’s face. Meta AI failed to realize that the selfie was fake and the account was successfully stolen.
After users complained about the vulnerability in Meta AI’s help system, Meta’s vice president of communications Andy Stone said the vulnerability had been patched and the affected accounts were restored. The incident once again shows the problems that arise when human customer service representatives are completely replaced by AI.
