Google has developed Mirror Dimension, a massive simulated internet designed to safely test AI agents without exposing real websites or users to potential risks.
Google has developed Mirror Dimension, a massive simulated internet designed to safely test AI agents without exposing real websites or users to potential risks.
Featuring thousands of synthetic websites that replicate shopping platforms, social networks, and online services, the environment evaluates Chrome's AI browsing agents as they complete complex tasks and defend against cyberattacks. Google says the system can generate more than 1,000 security test candidates per engineer daily, improving testing efficiency by over 30 times compared to manual methods.