A Green Run Is Not a Green Light: Engineering Responsibility in Browser Automati

<p>Most browser-agent demos still ask a capability question:</p> <blockquote> <p>Can the agent use the browser?</p> </blockquote> <p>That question made sense early on. A browser

<p>When an <strong>AI browser agent</strong> fails, the first instinct is often to rewrite the prompt.</p> <p>Make it clearer.</p> <p>Add more steps.</p> <p>Add more warnings.</p>

<p>Your <strong>AI browser agent</strong> can open a page.</p> <p>It can click buttons.</p> <p>It can fill forms.</p> <p>It can even complete a workflow that looks correct from

<p>You launch a headless Chrome instance, point it at a target site, and the first response is a 403 or a CAPTCHA. The IP is residential. The user-agent string looks like a normal