Overview
- An advisory titled "Cybersecurity Must Block AI Browsers for Now" recommends blocking AI browsers unless organizations complete rigorous risk assessments and implement strict controls.
- Gartner warns that agentic features can be manipulated through prompt injection, cause erroneous actions, expose credentials to phishing, trigger incorrect purchases, and leak sensitive data to cloud back-ends.
- The firm says even approved deployments will require enforced usage policies, limited agent privileges, and active monitoring of AI browser activity.
- Perplexity’s Comet and OpenAI’s ChatGPT Atlas are cited as examples of AI-powered browsers that can automate tasks and pose elevated risk in corporate environments.
- Earlier research from SquareX and Cato Networks, including the "HashJack" URL-based attack, underpins the advisory, while Indian regulators continue reviewing agentic AI risks and market impacts.