Remember 2014, when a U2 album quietly showed up in every iTunes library on earth whether you asked for it or not? Apple got roasted for that, apologized in its own stiff way, and shipped a removal tool. The lesson seemed to land: don’t put stuff on someone’s device that they didn’t choose.
Twelve years later, that lesson has apparently been filed under “revenue opportunity.” In 2026 Apple started running persistent ads inside iOS itself, pitching iCloud+ and AppleCare+ to people who already handed over a four-figure sum for the hardware. Users are not thrilled. One line making the rounds sums it up better than any analyst note: “I wish Apple would just stop that crap.”
Why a toolkit reviewer cares about someone else’s phone
I spend most of my time here testing AI tools, so a complaint thread about iPhone notifications might look off-topic. It isn’t. What happened to iOS is the exact failure mode I watch for in every product I review, just running on a much bigger stage.
The pattern goes like this. A tool earns your trust by being good. You build habits around it. You move your data in, your workflow in, your team in. And then, once leaving would cost you something real, the business model starts asking for more. Sometimes that’s a price hike. Sometimes it’s a feature moving behind a higher tier. And sometimes it’s ads, wedged into the surface you already paid for.
Apple’s version is unusually clean because the ads aren’t even selling you a third party’s product. They’re selling you more Apple. The pitch inside the phone is for the storage subscription and the extended warranty. It’s upsell as an operating system feature, and it’s part of a broader services push that has been the company’s growth story for years now. The hardware margin has a ceiling. Recurring revenue doesn’t.
The lock-in tax, stated plainly
The most honest reaction I’ve seen to all this came from a user explaining why they’re stuck: the only Apple thing they genuinely depend on is iMessage, and leaving would mean abandoning every group chat they’re in. Everything else has a Linux equivalent they’d happily switch to.
That’s the whole mechanic in two sentences. It isn’t that the product got so good you can’t live without it. It’s that one specific piece of it has your social graph hostage, so you’ll sit through the ads.
Swap “iMessage” for the AI tool of your choice and the sentence still works. The vector database holding two years of embeddings. The agent framework your whole pipeline is written against. The note app with your entire second brain in a proprietary format. The transcription service where your archive lives. None of those are good products by virtue of being hard to leave, but all of them get graded on a curve because of it.
What I actually check for now
This story changed my review checklist, so I’ll share the parts that matter:
- Who profits when I’m annoyed? If a tool makes money from my attention as well as my subscription, those two revenue lines will eventually fight, and attention usually wins.
- How much does exit cost? Not “is there an export button,” but: can I leave on a Tuesday afternoon with my data intact and my team functional? If the answer is no, treat every current perk as temporary.
- Where’s the growth pressure coming from? A company whose core business has plateaued will look for money somewhere, and the somewhere is almost always the installed base.
- Are paid tiers actually free of promotion? “Pay us and we’ll stop selling to you” used to be the deal. That assumption is no longer safe, and it’s worth confirming before you commit a workflow.
- Is there a credible alternative I could run myself? Not necessarily one I will run. One I could, if things go sideways.
The uncomfortable part
Nobody in the AI tool space has anything like Apple’s position, which is precisely why this is worth watching. Apple has more goodwill, more pricing power, and more reason to protect its reputation for taste than any startup shipping an agent wrapper this quarter. And it still decided that putting promotional material inside a paid operating system was worth the backlash. One forum comment described the company as having lost the plot in 2026, which reads harsh until you consider what it takes to make longtime customers say that out loud.
If the most disciplined product organization on the planet reaches for ads once growth gets difficult, your favorite AI tool — venture-funded, unprofitable, burning cash on inference — will reach for something similar and sooner. Maybe not ads. Maybe usage caps, maybe your data training the next model, maybe a “Pro” tier that quietly absorbs features you already use.
So pick tools you could walk away from. Not because you plan to, but because the option is the only use you’ll ever have. Apple just showed everyone what happens when customers don’t have it.
đź•’ Published: