SourceChrome Enterprise announced enhanced security capabilities and a three-pillar architecture to secure autonomous AI agents operating within the browser.
- Introduces real-time Data Loss Prevention (DLP) for AI agent workflows in Chrome Enterprise Premium.
- Implements User Alignment Critic inspired by Google DeepMind CaMeL research to veto malicious prompt injection hijacking.
- Extends Site Isolation to restrict AI agent execution strictly to task-relevant origins.
- Expands Vulnerability Rewards Program (VRP) up to $20,000 for verified browser agent security boundary breaches.