A new AI scam targeting 1.8 billion Gmail users has prompted Google to issue a “red alert.” Cybercriminals exploit Google’s Gemini app, an AI assistant, using a unique tactic described by tech expert Scott Polderman. Hackers send emails containing hidden messages that instruct Gemini to extract users’ login credentials without any visible prompts. This “AI against AI” scam could set a precedent for future cyber attacks. Users are cautioned that no links need to be clicked for the scam to activate; Gemini could notify users of potential risks. To safeguard against this vulnerability, security expert advice includes disabling Gemini’s features in Google Workspace and turning off “SMART FEATURES.” As generative AI adoption rises, Google warns of emerging threats like indirect prompt injections, which involve hidden malicious instructions in emails and documents. Google emphasizes its multilayered security approach to combat these sophisticated attacks. Users should remain vigilant and update their security practices accordingly.
Source link
Share
Read more