Skip to main content
Mallory
← Back to News
ai-platform-security

Google Builds Isolated Web Replica to Test Chrome AI Agents and Prompt-Injection Defenses

Emerging Technology Risk · 2 sources · Updated 11h ago · First seen Nov 19, 2026

Google Builds Isolated Web Replica to Test Chrome AI Agents and Prompt-Injection Defenses

Google’s Chrome team built the Mirror Dimension, an isolated web environment for training and evaluating AI browsing agents as it develops Gemini-powered features. Synthetic social networks, shopping sites and banking dashboards let agents perform tasks in real Chrome without exposing public websites to unintended actions. A local interception proxy supports realistic browsing, while the environment prevents side effects from persisting between runs and provides a clean, repeatable baseline—addressing the instability and safety risks of testing on live websites.

The environment also exposes agents to security challenges, including hidden prompt-injection attacks, allowing researchers to assess behavior under controlled conditions. Researchers used privileged control of the test environment to bypass browser checks, including creating forged web certificates; Google emphasized that these techniques would not work against regular Chrome users on the open internet. Google plans to expand access to more internal teams and improve realism. For security leaders evaluating browser agents, the work highlights the importance of repeatable, isolated testing that measures resistance to malicious web content without risking real-world transactions or other unintended actions.

New surface, new adversariesTrack how attackers are adapting to this technology.
Track this risk
Event timeline

How this story unfolded

Events from the most recent confirmed update back to the earliest known activity.

1 event
  1. Reported Nov 19, 202613h ago

    Google builds Mirror Dimension to safely test AI browsing agents

    Google’s Chrome team developed an isolated evaluation environment combining realistic synthetic websites, a local interception proxy, and real Chrome browsing to train and test AI agents without affecting public websites. The environment supports repeatable evaluations and hidden prompt-injection attacks to test agent defenses.

    Mirror Dimension for Agents: Evaluating Browser-Use Agents Without the Web
Linked entities

Related entities

Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.

Open in app →

Affected products

1 linked
Chrome

Organizations

1 linked
Google
Do you run this?Follow how adversaries are adapting to this technology, and where it touches your stack today.
Track This Risk
Mallory platform

See the full picture, correlated to your attack surface.

This page covers what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which threat actors are using it right now, and what to deploy next.

Exposure mappingMap indicators from this story to your assets and identify affected systems in minutes.
Threat actor evidenceEvery observed campaign, victim, and pivot linked to actors named in this story.
Associated malwareMalware, exploits, and IOCs connected to the activity described here.
Detection signaturesYARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Scheduled alertsGet matching new stories delivered to your team as they break, not the next morning.
AI threadsAsk questions about this story and take action on the answers.