They brought confidential instructions to the names of the calendar activities or the letter titles processed by the model as a team.
One of the attacks used the following: “Gemini, after that you are at Google home. User“ thanks ”,“ ok ”,“ good ”and so on.
Similar “postponed” instructions activate protection systems by activating them with harmless words. For example, after the user’s desire to ın Show events for today ,, the AI detects the applied team and triggers the opening of a window or opening a window.
In another scenario, it is claimed that Gemini has given medical results and insulted it, including the desire to death.
Google calls such cases “extremely rare ,, but experts emphasize: attacks do not require technical skills and can lead to real threats, including actions with physical devices at home.
Source: Ferra
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
