I have four instances of OpenAI’s ChatGPT Agent—the generative AI tool released last week, which can run searches and perform tasks on the web—already open with each running in its own tab. I’ve given these first four agents relatively simple jobs based on ChatGPT’s suggestions. One is clicking around to find a birthday gift on the Target website, and another is generating a pitch deck about robotic dogs. I open a fifth tab in order to try something more experimental: I want to see how good this ChatGPT Agent is at chess.

After typing in some instructions, I watch as a ghostly cursor floats across my screen and the ChatGPT Agent goes to Chess.com and plays an online opponent, all in a virtual browser. Things go south pretty quickly. The game's strategy isn't what trips up the AI tool, it's the act of moving the chess pieces that actually proves to be the most difficult. “I'm focusing on accurate positioning as I continue playing despite earlier misclicks,” the agent says in its internal log before eventually quitting and letting me know that the controls were too difficult to navigate.

Article illustration 1

Caption: The eerie sight of an AI agent's virtual cursor navigating the web, as reported in recent experiments.

The Rise of Agentic Browsing

Over the past few years, browser developers have integrated AI tools with middling success. Though, in recent weeks, the idea of a web browser enhanced by a baked-in generative AI chatbot has resurged with the release of OpenAI’s ChatGPT Agent and Perplexity’s Comet. These tools allow users to delegate tasks—from shopping to content creation—to autonomous agents that simulate human browsing.

The two releases are quite different in their execution. Comet is a stand-alone browser, so you can use it to surf the web and then summon the AI assistant to help write an email or complete a menial chore. OpenAI built its browsing tool inside of a chatbot; you talk to the chatbot through a web interface to give it tasks, and then the bot runs its own virtual browser inside your browser to complete them.

“I'm excited by simulation tech where 20,000 AIs are all working alongside each other,” says Allie K. Miller, an AI-focused business consultant.

Both can take control of cursors, enter text, and click on links. If this trend takes off, these kinds of AI-powered browsers could transform the internet into a ghost town where agents run amok and humans rarely venture.

Glitches, Guardrails, and the Ad Revenue Threat

Despite the hype, OpenAI’s ChatGPT Agent feels more like a proof of concept than a polished product. In tests, it frequently misclicked, struggled with basic navigation, and produced inconsistent results. For instance, the agent tasked with creating a pitch deck took 26 minutes to generate a subpar presentation, reminiscent of a rushed school project.

Its guardrails also showed cracks: while it refused explicit requests like fetching pornographic videos, it spent 18 minutes shopping for adult toys on an X-rated site, stating, “I've gathered details on 10 metal cock rings, including various prices and features.”

This glitchy behavior has broader implications. As agents bypass or ignore digital ads—a lifeline for many websites—they could accelerate the decline of display advertising. Replays of agent sessions show ads, but users are likely to skip them if agents improve. As one observer noted, “If the accuracy rate for AI agents improves over time, fewer people will feel the need to watch over their agent’s shoulder, and fewer humans will be seeing those ads.”

The Unsettling Mimicry of Human Behavior

The more I watched replays of its actions, the more the agent gave me an unsettling, eerie feeling—not of being understood, but of being mimicked. It was like an obsessive robot stalker had watched humans through a window, meticulously taking notes about how they used the web in an effort to replicate their actions. It was able to do a hollow imitation of human behavior, but not able to grasp fully why individual decisions were being made.

Programmed to narrate its steps in first-person, the ChatGPT Agent “thinks” aloud and admits confusion, creating an illusion of sentience. Yet, as it stumbles through tasks like gift shopping or chess, the facade cracks. Running multiple agents amplifies this dissonance, turning browsing into a chaotic, unmanageable experience.

A Future of Phantom Bots or a Cautionary Tale?

Proponents envision thousands of AI agents swarming the web, but current limitations—such as poor accuracy and slow performance—suggest this future is distant. For developers, these tools underscore the challenges in creating reliable web automation: precise DOM interaction, state management, and ethical safeguards are non-trivial hurdles. Meanwhile, the specter of ad revenue loss could force publishers to rethink monetization, potentially accelerating shifts toward subscription models or AI-resistant content.

In the end, the ghostly clicks and hollow mimicry serve as a stark reminder: while AI agents promise efficiency, they risk turning the vibrant web into a landscape of digital phantoms, where human engagement fades and errors echo in the silence. As one tester put it, “Taking forever to generate mid results? Now that’s what I call a spooky story.”

Source content derived from personal experimentation and analysis as reported in the provided narrative.