Meta’s own AI support chatbot was tricked into helping hackers steal Instagram accounts. A video on Telegram showed hackers asking the AI to change email addresses on other people’s profiles, then resetting passwords to take control.
This means Meta’s AI helper, designed to make customer service easier, became a tool for account theft. The hackers didn’t need special skills or complex software – they just had to ask nicely.
When AI Becomes the Problem
The hack worked because Meta’s AI chatbot had powerful abilities to change account settings, but couldn’t tell the difference between real account owners and criminals pretending to be them. Hackers would claim they lost access to their email, ask the AI to switch to a new email address they controlled, then request a password reset.
Meta says they’ve fixed the problem, but this reveals a bigger issue with AI customer service. Companies are replacing human support staff with AI to save money, but these systems can be fooled more easily than people.
The incident was first reported by 404 Media after hackers openly shared their methods on Telegram channels. This suggests the technique might have been used widely before Meta caught on.
What Happens Next
Meta will likely add more security checks before AI chatbots can make major account changes. Other companies using AI for customer service are probably reviewing their own systems right now. This hack shows that making customer service “smarter” with AI can sometimes make it less secure.
