A lesson in separation of concerns: Google’s gemini hijacked

The news Google’s Gemini AI was hijacked via a hidden prompt in a calendar invite.

Hackers Hijacked Google’s Gemini AI With a Poisoned Calendar Invite to Take Over a Smart Home. For likely the first time ever, security researchers have shown how AI can be hacked to create real-world havoc, allowing them to turn off lights, open smart shutters, and more.

Rings bells to me about the lack of human data interaction especially the separations of concerns.

I already talked about this with Google’s AI wanting access to everything Google related. This hijack wouldn’t be so effective if there were better permissions and separations of concerns were built in rather than given lip service.

Google sent an email today with the title… Introducing temporary chats and new data controls. Still not addressing the underlying problems.

When will they learn???

It really makes me think about a few things.

Will Android fork off sooner or later? Will moat people care?