“Google is a monopolist, and it has acted as one to maintain its monopoly[…]in general search services and search text advertising through a web of anticompetitive practices”, wrote Judge Amit Mehta back in August 2024. That line set the tone for what was supposed to be the biggest antitrust reckoning in tech since Microsoft in the 90s.

And so it started the evaluation of some remedies proposed by the U.S. Department of Justice, that we covered extensively in Artifacts last year.

They included what may read like a Silicon Valley nightmare: force Google to sell Chrome, tear up its default search deal with Apple, maybe even split off Android.

This could’ve been HUGE. But it wasn’t, eventually.

A couple of days ago, the Judge Mehta published a 223-page case with all the remedies significantly watered down:

• No Chrome sale. According to Mehta, breaking it off wouldn’t restore competition, and in any case it wouldn’t feel “natural.” Alphabet keeps its browser, which happens to be the gateway for 7 out of 10 internet users (with Google as the default search engine)

• No end to default search deals. Apple will keep cashing Google’s checks (20 billion dollars a year, for the record) and Google will keep getting all that iPhone traffic.

• The one real concession: Google must share parts of its search index data with rivals like DuckDuckGo, Yahoo, OpenAI, and Perplexity.

But really, the reason things didn’t go further is external. There’s a new elephant in the room - two letters: AI.

Mehta put it clearly: this case started in a different (search) world: back when “search” meant typing words into a box and getting ten blue links.

Today, AI has scrambled the field. The remedies that made sense in 2020 would be useless in 2025. You don’t fix yesterday’s monopoly if tomorrow’s competition looks completely different.

And therefore,

You need “humility” as you cannot “gaze into a crystal ball”.

And as the issues you’re dealing with changed in a direction you cannot forecast, the remedies need to change as well.

Here’s the real takeaway: one of tech’s landmark antitrust cases ended, not with fireworks, but because AI stepped onto the field. Suddenly, the competition is between a whole new cast of players.

And if we’re to give AI search a shot, as I wrote back in November 2024 😇, “Piecemeal solutions won’t be enough. Structural remedies are needed.”

That’s why the meaty remedies - Chrome, Android, Apple deals - vanished. The only thing left standing is data sharing, a bet that competition in the AI search era won’t be about who has the best lawyers or the biggest checks to Apple, but who has the fuel to train models.

What data? Google will need to share parts of its search index, the corpus of web pages and information that feeds its results page. But this does not include other data associated with those results, including information about the quality of web pages.

Why data? Because it might finally level the field. Smaller rivals like Perplexity or DuckDuckGo have the smarts but not the raw material. Without access to a real index of the web, they’ll never catch up to Google’s decades of crawling and ranking. With some extra fuel, they just might.

The DOJ put it bluntly: a newcomer can probably build a search system that answers 80% of queries in the short run. It’s the remaining 20% - the weird, hyper-specific, often forgotten corners of the internet - that Google has locked down.

Sharing data is supposed to fill that gap. Maybe.

This way, the competition should shift from who has, inevitably, the best data (Google) to who’s best at AI.

The playing field hasn’t changed much - it’s still about who can pull quality search data and deliver it to users. The pros call this the “ecosystem of information retrieval.” Now, though, there’s a new player you can’t ignore.

For AI to succeed at parsing, ingesting, and refactoring information, it needs more than just smarter algorithms.

If AI is ever going to become a normal part of how information gets distributed, it has to reach a level where it can actually compete.

All of this is through data, and a remedy that acknowledges that, regardless of AI potential, its adoption will make it a thing, for real.

And that’s where antitrust comes in: it’s less about fixing the past than making sure tomorrow is open - whatever it ends up looking like.

Will sharing data be enough to make information retrieval different? Too soon to say.

Save for Later

Quite interesting to see how people are using ChatGPT, which is getting safer btw! And, well, there’ve been some safety issues.

We may stop handwriting, but at least ghosting should be forgiven!

Speaking of Google, you can now pick and choose your sources. Speaking of sources, why Wikipedia is still so good.

What if tech regulation goes back to school? Someone is doing the math of AI energy consumption :)

Where next?

Artifacts will be covering Italian Tech Week in Turin from October 1st to 3rd. The agenda looks great - plenty of top speakers and a wide mix of topics!

And a few more events where Artifacts will be in as a media partner!

• October 4: Bayram, Rome 🇮🇹

• November 5–7: Mozilla Festival, Barcelona 🇪🇸

• January 23–24: Political Tech Summit, Berlin 🇩🇪

If you’ll be there, just hit reply to this email and let’s meet up!

The Bookshelf

If you’re curious about how our search for information has evolved - from oral storytelling all the way to Google and AI - “The Information” is a solid pick. It’s a deep dive, best suited for the nerds among us, but worth the read.

📚 All the books I’ve read and recommended in Artifacts are here.

Nerding

If you use a Mac, this is a game changer. Everything is a keystroke away and shortcuts are so good. It’s like Spotlight on steroids. If you use a Mac, you’ll never go back from Raycast.

☕?

If you want to know more about Artifacts, where it all started, or just want to connect...

Let's grab a virtual coffee ☕