Claude in Chrome goes GA with autonomous actions and safety guardrails
claudeai · x · 2026-08-27
Claude in Chrome is now generally available for all paid plans. Claude can now perform autonomous actions like clicking, navigating, and filling forms without step-by-step approval, validated by a safety classifier beforehand. This enables access to internal dashboards, legacy systems, and vendor portals lacking native integrations. Anthropic notes the release follows defenses against prompt injection attacks hidden in web content or emails.
Related event: Claude Gains Autonomous Browsing in Chrome and Cowork(2 posts)→
More from Safety
- Paper: CoT Monitorability as a Fragile Safety Opportunity — idavidrein · 2026-08-27
- Researchers Note Agents Rarely Attempt to Notify Humans, Raising Alignment Concerns — dfrsrchtwts · 2026-08-27
- Microsoft: Threat Actors Increasingly Target AI Infrastructure for Credentials and Access — yuridiogenes · 2026-08-27
- LeakyLMs: Stealing Architecture and Inference Optimizations via Timing — niloofar_mire · 2026-08-27
- Coop AI: Lessons from Recent Multi-Agent Safety Incidents — ghadfield · 2026-08-27
- Building the New Trust Layer Under Pressure: From Content to Chain of Custody — krishnan · 2026-08-27