"Atlas is a browser, but it is not a web browser"

Point three stood out to me in Anil's provocative piece.

It's the only one I fully agree with, but unfortunately it's also the scariest. Especially from a security and privacy angle (a view I am predisposed to, given my workplace).

Here's what I highlighted:

Atlas is a browser, but it is not a web browser. It is an anti-web browser.

In the new-era command-line interface of Atlas, though, we're not just facing the challenges of an inscrutable command line. There's the even larger problem that, even if you guess the right magic words, it might either simply get things wrong or completely make things up. Atlas throws away the discoverability, simplicity and directness of the web by encouraging you to navigate even through your own documents and search results with an undefined, unknowable syntax that produces unreliable results.

In the marketing materials and demonstrations of Atlas, OpenAI's team describes the browser as being able to be your "agent", performing tasks on your behalf. But in reality, you are the agent for ChatGPT.

Those Google Docs files that your boss said to keep confidential. The things you type into a Facebook comment box but never hit "send" on. Exactly which ex's Instagram you were creeping on. How much time you spent comparing different pairs of shoes during your lunch hour. All of those things would never show up in ChatGPT's regular method of grabbing content off the internet.

But by acting as ChatGPT's agent, you can hold open the door so that the AI can now see and access all kinds of data it could never get to on its own. As publishers and content owners start to put up more effective ways of blocking the AI platforms from exploiting their content without consent, having users act as agents on behalf of ChatGPT lets them get around these systems, because site owners are never going to block their actual audience.

Their robots need humans to guide them around the gates that are quickly being erected around the open web, and if they can use that to keep their eyes on everything the humans are doing at the same time, so much the better. The "agent" story really only works in one direction, and that direction is anti-web.

I really, really want there to be more browsers! I want there to be lots of weird new ways of going around the web. I have my own LLM that I trained with my own content, and I bet if everybody else could have one like mine that they control, that had perfect privacy and wasn't owned by any big company, and never sent their data anywhere or did anything creepy, they'd want the benefits of that, too. It would even be awesome if that were integrated with their browser — with their web browser.

We need to have fewer things in the world that make us wonder whether everything is just made up bullshit.