Seems they just want your personal browsing history for training, and built this Chromium reskin to get around privacy controls.
> "OpenAI decided to build its own browser, rather than simply a "plug-in" on top of another company's browser, in order to have more control over the data it can collect, one source said."
I think native browser can provide better integration than browser extensions. Things like: mcp, access to accessibility, playwright, browser use, stagehand - this should help make a better operator.
> The browser’s access to a user’s web activity would make it the ideal platform for AI “agents” that can take actions on their behalf, like booking reservations or filling out forms, directly within the websites they use.
How about agents canceling services I no longer want? Or agents figuring out what choices I have to make to get my return picked up rather than have to take it to a drop off or pay?
There was the open web, with lots of rich api's... All of that got killed for the sake of profit and market consolidation. Are those same players going to be tolerant of agents cutting into their bottom line, of agents that cant be upsold, that dont misclick?
IF we ever get a solid agenic tooling I suspect that it will be murdered in its crib by industry.
Why do these idiot companies keep tackling real technical challenges that they don't want.
Open AI doesn't want to own and maintain a browser for the next 20 years! They'll give up in like two years when they realize it's actually quite hard and they have no particular committment to it
Not really if you are a Chromium fork. Difficult from ground-up but from Chromium it’s rather easy. Way easier than creating an LLM. Even easier when you can hire the Chromium developers.
You clearly think how they think. It's just so EASY to do. Practically done already. With an attitude like that how could they not succeed in tackling a really, really hard problem that requires nearly endless thankless investment?
If it were rocket science it would be less surprising to me that they would want to be spending their time on it.
It's clear that they envy Google's position of owning a web browser (that everyone uses). But a play like this that involves entering a highly competitive market with openly vicious level of scrutiny... Like, what would you say if Facebook announced that they were announcing that they would be starting their first hospital? That's a hypothetical, but IRL Amazon tried to convince people that they wanted to be operating a chain of hair salons. My point is that it was easy to tell from the outside that this was corporate whim and they weren't going to be very willing to put up with it for long if it turned out to be genuinely hard, because there just wasn't enough alignment with the mission.
So yeah, I've no trouble understanding why they want to own a browser in theory. And since they can print money and want to hoover up talent anyway, why not just hire some engs. It looks good for the company's future growth prospects if people do believe you're serious. But it just seems to me that thus far they've shown no real hard commitment, no philosophical alignment, and no appreciation for the reasons what they're doing is difficult: it's a social problem not just a hiring one
It's actually super easy. I modified Chromium's internal code myself to do what I wanted, I added some logic in C++ and then compiled it. It just works. They mainly want to gather more data and have access to more functionality, so the LLM agent isn't constrained by the plugin API etc. It's not really hard. They won't do any browser level stuff, instead they'll basically create their own plugin on top of Chromium with full access to everything, and that's it.
Like The Browser Company, that gave up on Arc to make Dia, an "AI browser" that's sure to have an equally short life.
I get that everyone wants to be a platform and to live where the user is all day, but getting folks to switch browsers is nearly impossible and, on Apple, fully impossible as only Safari-based browsers are permitted and those are highly restricted.
At first I thought this was funny, but truthfully I don't love my browser. Brave and Firefox are both mildly annoying in different ways.
I'll probably give it a spin when they release it. It's just a Chromium fork and they can afford to make it good, if the AI integration is subtle and actually useful and there are robust controls on how my data is used (and it supports high quality adblock), I could see myself using it regularly.
We're on a Google and OpenAI collision timeline. Google has to win to stay relevant, while OpenAI has to win to survive.
Looking at my own usage patterns, Google's ad business should have be long gone now, but then again I've been running adblockers since forever so maybe I just don't understand the dynamics here. Google has to recapture the perceived (or actual) lead in the AI market before its made obsolete and OpenAI has to keep Google on the defense otherwise it will just be commoditized away (I think this is more about good old market share here and no one waits around for AGI to unlock a spectacular economic value, they want each other's money now).
This is the second step in just a month or so where OpenAI directly encroaches on Google territory (first was with the recruitment of Johnny Ive to sidestep established mobile platforms and now going straight for Google's jugular).
It would be really interesting once OpenAI gets into the ad business, because right now Google has its share of ad revenue on hard allocation, and once in-stream AI advertising opens up it would surely divert some of that money.
Almost certainly a Chromium fork rather than just a layer, similar to how Edge, Brave and Arc are built. This gives them full control over the rendering engine and browser internals while leveraging Chromium's compatibility.
This would make giving them all my data so much easier for them. That is the endgame, right? OpenAI becomes the advertiser of record using the best, richest, most targeted data ever. And we gave it all to them. “Dumb Fucks” -Zuck
Why not just release this as an extension? And make it work with Firefox as well. This seems why browser extensions exist at all. No need to pretend that perplexity or openai is actually going to do any work at all on the rest of the browser. And it seems overkill to release your own browser just to white label it.
IIRC most web use is mobile. Approximately 0% of mobile users can install extensions.
Edit: I forgot iOS Safari can install extensions. Android users can install Firefox and then install extensions, but this is a big ask for the typical user.
Probably still a good estimate that <0.5% of mobile users have ever installed an extension.
I do it for Ublock Origin. The web without an adblocker sucks so much ass. I truly don't understand how regular users do it without killing themselves.
If that is a literally true, I'd be doubting her capability to informed consent to a relationship, but that's just me. It's literally a google search and one or two button clicks. It's about the same complexity as opening and scrolling Facebook.
> OpenAI decided to build its own browser, rather than simply a “plug-in” on top of another company’s browser, in order to have more control over the data it can collect, one source said.
It's pretty much impossible to create WebExtensions that provide a complete experience and feel native. The API is (perhaps rightfully so) quite restrictive.
> "OpenAI decided to build its own browser, rather than simply a "plug-in" on top of another company's browser, in order to have more control over the data it can collect, one source said."