Johannes von Cramon

John Gruber argues that no AI company will use the Android AI interoperability APIs the European Commission just mandated for Google, just as no 3rd party browser used Apple’s DMA browser engine APIs. His reason: the EU user base is not large enough to justify building a custom web browser, or an AI assistant.

But market size is not the only reason. According to Open Web Advocacy, Apple’s entitlement terms were one-sided and failed the DMA’s own standard of strictly necessary and proportionate. Vendors had to ship a brand-new app for the EU and abandon their entire existing EU user base to do so, which is not commercially viable.

Everything will be determined by Google’s terms.

There was a lot of discussion this week about cheaper Chinese AI models after the release of the new Kimi K3 model. The tech media framed it primarily as the beginning of a price war. That framing seems off.

Ben Thompson’s “Who’s Afraid of Chinese Models?" piece rightfully argues that the correct result matters more than the token. Models burn different amounts of chain-of-thought tokens to reach the same conclusion.

And there might be another reason for the price difference:

Right now there is a price umbrella that is downstream of the lack of compute; I highly doubt that Chinese models are cheaper to serve on a marginal cost basis, they just seem cheaper because Anthropic and OpenAI are so supply constrained that they are charging far more than they would if there were sufficient supply to meet the demand for intelligence.

The researchers Agarwal and Sen ran an interesting field study with 1,065 US Chrome users in January this year. The users had to install an extension toggling AI Overviews off and on. The paper is the first randomized evidence on what this actually means for publishers.

Google’s standing defense has been that AIOs reduce clicks but improve quality of the remaining ones, filtering out low-intent traffic and educating the user with better search results beforehand.

Well…

On queries where AIOs appeared, outbound clicks fell 39.8%. Across all searches, the drop was 18.5%. When AIOs were removed, the recovered clicks showed no quality difference: bounce rates and time on page were unchanged. The traffic from Google was nothing more than ordinary traffic.

Also worth noting: Links within AIOs generated only 7.1% of outbound clicks. Very telling for everyone who is betting on GEO as the next big thing in search marketing.

More than 60 apps passed Apple App Store review while hiding an online gambling platform behind weather icons, maps, and travel games. The front end behaved correctly for reviewers, then switched to a betting UI the moment a Brazilian IP address connected.

As if that weren’t enough, the operators published a playbook on GitHub explaining how to vibecode these apps using a Cursor agent, making the operation look like the work of dozens of independent developers.

But it gets better: Apple’s “You Might Also Like” recommendation engine clustered these apps together and even surfaced them during discovery, grouping scam apps as if they were a genre.

What a massive L for the Apple review team.

On July 14, Apple notified advertisers about an update to its Terms of Service. The old text defined ad properties as covering “Apple software applications or Apple devices.” The new text adds “web applications, or other platforms or properties” and drops the word “Apple.”

Aside from the obvious rumors of an iAd 2.0 network, this changes Apple’s two-tier attribution regime since ATT. Until now, Apple used its own Ads Attribution API for its own inventory (read: ASA) and SKAdNetwork/AAK for everyone else. That asymmetry held while Apple Ads remained on Apple-owned surfaces—App Store, Apple News, Stocks, and Apple TV— with Maps ads arriving this summer.

It collapses the moment Apple Ads appear on a surface Apple doesn’t own. Apple would measure its own ads on your app better than you can measure your own.

Earlier this month, I laid out the contract math on Netflix renewals. Since the costs of casting and showrunning significantly increase from season to season, they prefer producing low-budget, limited-run series.

Eric Seufert did a great follow up with a piece that adds the demand side: Netflix' recommendation engine runs the same calculation. Routing a user to a new series extracts more economic value than returning them to a known one:

A fan who loved Season 1 comes back for Season 2 on their own, so that slot mostly buys a view Netflix would have gotten anyway, whereas the same slot spent on a new title surfaces something the viewer would not have found alone and spreads their attachment across more of the catalog, which means higher incremental engagement and lower churn risk.

A New York Times analysis published this week adds data to a feeling that I believe many of us already had: over the past decade, US productivity grew at roughly twice the rate of real compensation. Labor’s share of national income sits near a 78-year low. Tech employment contracted for 18 consecutive months. Finance shed more than 100,000 jobs since May 2025. Oil and gas went from roughly 200,000 workers in 2013 to about 115,000 today.

AI advocates tend to skip this step. They cite efficiency, output per hour, leaner operations. While that might be correct, the other side of the coin is fewer people doing more, with the margin flowing to balance sheets rather than paychecks.

Wild stat from a recent Marketing Science study: 96% of sponsored tweets carried no disclosure. Other studies found similar rates on YouTube and Pinterest.

To this day, the FTC has never obtained a monetary penalty from an individual influencer for a disclosure violation. The agency mostly targets brands, because policing millions of individual posts is genuinely hard.

And as you’d expect, Generative AI makes this bad situation even worse, with campaign costs approaching zero and fake influencer avatars promoting real products.

In the early days of Growfirst, an app marketing agency I co-founded in 2016, we constantly cited Hopper as a best practice for cutting-edge growth marketing (or growth hacking, as it was called back then). Highly relevant push notifications, perfectly timed permission prompts, and gamification that kept users engaged even when the app’s use case (flight booking) was infrequent.

Then every screen became a nudge. The app grew genuinely unusable because of popups, re-engagement prompts, and ever-new monetization attempts.

Last week’s FTC $35 million settlement is therefore not a big surprise. Companies that treat optimization as the only value follow a predictable arc: growth → saturation → extraction. Hopper followed that trajectory precisely. The gold-standard case study became the cautionary one.

Ryan Broderick’s piece this week frames Netflix’s catalog strategy as deliberate shovelware production, and a Bloomberg report confirms it: Beef’s viewership dropped from roughly 30 million views to 10 million between seasons. Across big titles, the range runs from 30% to over 70%. Netflix describes this as audience behavior. The structure of their contracts explains it more directly.

Cast and showrunner contracts carry built-in raises that spike around the third season, and because Netflix owns its shows outright with no syndication windfall to recoup them, those rising costs never get paid back. Launching a new show resets the acquisition hook, surfaces fresh content for non-subscribers, and avoids the renewal premium.

The platform’s current biggest title is the 13th Harlan Coben adaptation, I Will Find You, a limited series that premiered June 18. Coben adaptations fit the model precisely: cheap to develop, internationally distributable, finite by design. No second-season audience collapse if there is no second season.