AI browsing now available for early testing in Brave
Summary
Brave announces an opt-in AI browsing feature for early testing, detailing security mitigations like isolated profiles and reasoning-based defenses to combat prompt injection and unintended actions.
Key quotes
Agentic browsing is also inherently dangerous. Giving an AI control of your browsing experience could expose personal data, or allow agentic AI to take unintended actions.
AI browsing in Brave never trains on your data, unlike other agentic browsers.
Brave’s AI browsing therefore keeps its storage separate from your regular profile: cookies, logged-in state, caches, and other site data do not cross profiles.
The post outlines Brave’s cautious rollout of agentic browsing via the Leo AI assistant. It emphasizes a multi-layered security approach to prevent malicious prompt injections and model hallucinations.