Your users want the off switch. Are you shipping one?

Something is showing up across very different products right now. Parents want a brake on teen accounts. People want ad-free, algorithm-free social. Others are deleting apps or downloading tools to block the parts that hook them. And regular folks are packing into library rooms to learn how to turn AI off. Different problems, one signal: people want control back. Let me catch you up on what that means for what you build next.

The controls parents asked for, twice

Two big platforms shipped parental tools in the same week. Twitch now lets guardians link a teen's account, block livestreaming entirely, kill private messages, set daily time limits across devices, and filter content by label. Meta brought supervision to Threads, a platform with 500 million monthly users, three years after launch.

Notice the timing on Threads. Meta added these controls under real pressure, facing lawsuits over child safety and heat from regulators. The feature is good. The lateness is the tell. Controls that arrive only after a lawsuit read as damage control, not care. If your product touches minors, the default settings you ship today are the ones a regulator reads back to you later.

People are paying to block your best features

Here is the uncomfortable part. The features your growth team optimizes are the exact ones users are now paying and fiddling to escape. The Switchly app lets people block Instagram reels, the Explore page, and comments, plus YouTube Shorts and the recommendation tab. One writer got so stuck doomscrolling he deleted Instagram and used the website only.

Others want a nudge instead of a wall. MeBeMe sends small missions tied to who you want to become, a reader, a runner, rather than locking you out. It stores all data on the device, needs no account, runs no ads. When the thing people crave is a break from your engagement loop, your retention metric and your user's goal are pulling in opposite directions.

The bet that the loop itself is the problem

Some builders are skipping the fix and rethinking the model. Yope raised $12.3 million for a social network with no algorithm, no ads, and no public content. Accounts are private by default. Money comes from subscriptions, not attention.

The founders ran over 100,000 interviews and found about 30% of young people already use "spam" or photo-dump accounts to share candid stuff with a small circle. Many share nothing public at all. Yope now has nearly 15 million registered users, and it claims more than half open the app five days a week. The read: people still want to connect, they just don't want an audience or a feed grading them.

When "off" is the killer feature

Privacy is turning into a product decision, not a legal footnote. Euki, a period tracker, stores data on your phone, needs no account, and offers a fake PIN that hides your data. The Mozilla Foundation audited six top period apps, found serious flaws in five including Flo and Clue, and said it strongly recommends Euki for anyone who puts privacy first.

Compare that to OpenAI. Beth Skwarecki flagged that ChatGPT Health quietly narrowed its privacy promise: now the only way to get protected "Health conversation" status is to hand over your connected health records. Storing less can be the feature. Asking for more can be the risk.

The deep cut

The sharpest signal is the library workshop. In Philadelphia, a librarian's "Avoiding AI" event drew a full room and over 2,000 likes on a post that usually gets a few dozen. In Maine, Hannah Cyrus had to cap registration at 30 and open a waitlist. These people are not anti-technology. They scan old documents with OCR and cheer AI for medical breakthroughs. What they hate is AI they did not ask for, turned on by default, hard to turn off.

That is the line for your roadmap. When you add an AI feature, if the design "seems to force adoption," as Cyrus put it, you are creating the exact frustration that fills those library rooms. Ship the feature. Ship the off switch next to it. Make it findable, not buried under three menus. The teams that treat user control as a feature will keep trust. The ones that treat it as friction to remove will find their users at a library learning to route around them.

Three questions for your team

• For your newest AI or engagement feature, can a user find and flip the off switch in under 30 seconds? If not, who owns fixing that before the next release?
• If a regulator or reporter read your default settings for minors out loud tomorrow, would you defend them or scramble? Name the one default you'd change now.
• Where do your retention metrics reward the exact behavior your users are paying other apps to block? Pick one, and decide if you're willing to trade a little engagement for trust.