China's kid mode covers the whole phone. Ours stops at the app.

You set the controls properly. Downtime on the tablet, your approval needed before anything installs, YouTube on the restricted setting, most of a Sunday gone. Then your ten year old asks a chatbot how photosynthesis works, and you find out it happened in a browser tab, inside an app you vetted two years ago, on the device you thought was handled.

Nothing broke. Every switch did what you set it to do. They were just set one layer too low.

The list of apps will always be one step behind

Look at where AI actually reaches a child now. It sits in the search results before they click anything. It is a tab. It is a feature that turned up inside an app you already approved, delivered by an update nobody read. Very little of it arrives as a new download with a new icon, which is the one moment most parental controls are built to notice.

So the parent keeps a list. A toggle in this app, a different toggle in that one, and next month a third has quietly added a chat button. The list is never finished, and the child only has to find the item you missed.

China moved the switch underneath the apps

This is where the comparison earns its place, as long as we separate the design from the politics.

In November 2024 the Cyberspace Administration of China issued guidelines telling three groups to build the same thing together: phone makers, app developers and app stores. The result, which the regulator calls minor mode and pushed out in April 2025, is one setting that travels. Turn it on and the device, the store and the apps are meant to switch state together, in what the CAC calls three-party collaboration.

The defaults are unusually concrete. As The Register reported when the guidelines landed, the mode caps daily use at an hour for children under 16 and two hours for 16 to 18 year olds, prompts a break after 30 minutes of continuous use, and blocks non-essential apps between 10pm and 6am. Once the daily allowance is gone, all but “necessary” apps and the exemptions a parent has chosen are suspended by default. And as the American think tank ITIF notes, “all settings require parental verification to change or remove”.

Read the other half of the same rules, though, and this becomes a design to borrow rather than a policy to admire. ITIF also records that content inside minor mode is expected to promote “core socialist values” and to cultivate “minors’ patriotism”. A switch that reaches every app on a phone is a powerful object, and in China the state has a view about what belongs on the other side of it.

The AI part came later. Under the interim measures for AI human-like interaction services, in force in China since 15 July 2026, providers have to identify minor users and give them a minor mode with, as the law firm Latham & Watkins summarises it, “personalized safety settings such as periodic reminders of reality and usage time limits”. Guardian consent is required below 14, and parents can block particular characters and cap spending. Because the container already existed, the AI rules plugged straight into it.

Here, the work sits with the household

Singapore regulates one of those three layers, and is unusually honest about the limits of that. IMDA’s Code of Practice for Online Safety for App Distribution Services, in effect since 31 March 2025, requires designated app stores to work out which users are under 18 and keep them from the highest age-rated apps. Asked about it in Parliament, MDDI said the quiet thing out loud: “Regulations alone are unlikely to prevent a parent or any adult from helping a child circumvent the age assurance measures.” We looked at that gate more closely in app store age checks and AI.

Which leaves the household. MDDI’s Digital Parenting Survey, fielded in February 2025 with 1,986 Singapore parents of children aged 2 to 17, found that only 37 per cent felt confident guiding their child’s digital habits. Exposure to inappropriate content was the top worry at 81 per cent. The reasons parents gave for struggling are the useful part: not enough time, a child reluctant to follow the rules, a child able to get round the controls, and their own limited knowledge of the tools.

That last one is fixable in an evening.

Three settings that sit above the app

On an iPhone or iPad, the layer you want is Screen Time, then Content and Privacy Restrictions. Inside it, Web Content offers “Unrestricted”, “Limit Adult Websites” and “Only Approved Websites”. That third option is the nearest thing a family has to the whitelist logic, and it matters because a chatbot in a browser is just a website. The same screen now also carries restrictions for Apple’s own AI, covering image creation, writing tools and intelligence extensions.

On Android and Chromebooks, Family Link holds the equivalent. Its website settings are “Allow all sites”, “Try to block explicit sites” and “Only allow approved sites”, with Google’s own caveat attached: “No filter is perfect.” Note the boundary, because it is the same problem: Google says website permissions cannot be managed on a child’s iPhone or iPad, and may not yet apply on Windows, Mac or Linux computers. Family Link also carries a plain on and off switch for AI, under Controls, then Gemini, then Gemini Apps.

The browser filter, the AI switch and the install approval sit under everything else, which is why they hold for a year rather than a week. Our own view, and we build AI mentors for children so we have an interest here: the browser filter is the one most families skip and the one that does the most work.

Then tell your child which switch you set

The survey found children who resist the rules and children who get round the controls. Both are easier when the rule is not a secret.

Google’s own advice to parents whose children use Gemini is worth stealing, because it is written for a nine year old rather than a regulator. Help them understand that Gemini “isn’t a person and cannot think for itself or feel emotions”. Tell them not to share their address, their school or family details. Warn them that it “can hallucinate and present inaccurate information as factual”, and that “our filters try to limit inappropriate content from appearing, but they’re not perfect”.

A parent who has said all of that has done something no device setting can. The settings buy you the evenings. The conversation is what survives the child growing out of them.