Hackers can read private AI-assistant chats even though they’re encrypted

AI assistants have been around for over a year and have access to our most private information, including sensitive discussions about pregnancy, divorce, and trade secrets. Providers of these AI-powered chat services encrypt these interactions to prevent snooping. However, researchers have found a vulnerability in significant AI assistants, except for Google Gemini, that allows someone monitoring the data packets to decipher responses with surprising accuracy. With this attack, the specific topic of 55% of captured responses can be inferred with high word accuracy, and 29% of reactions can be deduced with perfect word accuracy.

Source:ARS Technica

Leave a Reply

Your email address will not be published. Required fields are marked *