An AI-enabled browser is a web browser that embeds an assistant or agent to summarize pages, answer questions, automate tasks, or act on behalf of the user. Unlike a normal browser, it combines page rendering, extension code, user data, and model-driven actions in one workflow. That makes it more capable, but also expands the trust boundary.
In security terms, the risk is that the assistant may rely on browser context that can be influenced by extensions, injected page content, or other local components. If an attacker gains a malicious extension foothold, they may steer the assistant toward leaking data, following unwanted instructions, or performing browser actions the user did not intend. Defenses focus on tight extension permissions, least-privilege access to sensitive sites, separation between page data and assistant prompts, and testing whether AI features can be manipulated through scripts or content injection.