agentic browser

A web browser that integrates a large language modellarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. assistant with access to the user's tabs, page content and logged-in sessions, and that can act as a browser agentbrowser agent, n.: An AI agent that uses a web browser on your behalf — searching, clicking links, filling in forms and reading pages — to complete tasks such as booking, shopping or gathering information. — navigating, clicking, typing and completing multi-step tasks such as comparing products, filling forms or triaging email — on the user's instruction. Typical features are a sidebar chat that sees the current page, an agent mode that drives the browser while the user watches or works elsewhere, and memory of past browsing. The category took shape in 2025 with Perplexity's Comet (July), The Browser Company's Dia, Opera Neon, Copilot Mode in Microsoft Edge, Gemini in Google Chrome and OpenAI's ChatGPT Atlas (October). Because the agent reads untrusted web content while holding the user's credentials, indirect prompt injectionindirect prompt injection, n.: A prompt-injection attack in which the trick instructions are hidden in content the AI reads on its own — a web page, an email, a file — rather than typed by the person using it. — instructions hidden in a page that hijack the agent — is the defining security risk, demonstrated against several products in 2025. Defenses include asking for confirmation before sensitive actions, logged-out modes, limiting the agent to approved sites and injection classifiers.

I asked the agentic browser to find a cheaper flight, and it stopped at the payment page for my approval.