Google has unveiled a dual-model security architecture for Chrome to block prompt injection, directly addressing Gartner's call to ban AI browsers.