You have two companion apps open in the store, you have read both descriptions, and they say roughly the same things. The one part of either listing the developer did not write as marketing is the panel further down: the AI girlfriend app privacy label, where the store obliges them to declare what the app collects and who else receives it. It is the most useful thing you can read before you install anything, and also a self-declaration with four documented gaps. Knowing both halves is what makes the two minutes worth spending.
What an AI girlfriend app privacy label actually says
The two stores structure it differently. On the App Store, the panel on a product page sorts data types by whether they are linked to your identity and whether they are used to track you. Apple defines tracking narrowly: your app data combined with data gathered from other companies' apps, websites or offline sources, then used for advertising or handed to a data broker. Apple is also plain about where the information comes from — its guidance on privacy information on the App Store says the program is one where the information is self-reported by the developer.
On Google Play, the Data safety section runs longer. It lists the data types collected, the data types shared with third parties, the purpose of each — app functionality, analytics, advertising, fraud prevention — whether a given collection is optional, whether data is encrypted in transit, and how you can ask for your data to be deleted. Google says the same thing Apple does in its own words: the information is declared by developers and not verified by Google.
For a companion app, one line matters more than the rest, and it is not the long list of what is collected. Conversations are the product, so of course messages are collected; the app cannot answer you otherwise. The line to read twice is what is shared, and for which purpose.
Four things the label is allowed to leave out
None of these is a loophole anyone is sneaking through. They are written into how the schemes work, and a label can be entirely truthful while staying silent on all four.
Data that never leaves, or never lands. Google does not require some data to be disclosed as collected or shared at all — among other cases, data processed only on your device, data processed ephemerally, and data that is end-to-end encrypted. A blank where you expected an entry can mean the handling genuinely qualifies, rather than that nothing happens.
The partners behind the categories. Apple's page notes that a developer or its third-party partners may use collected data for purposes such as advertising or analytics. The label names data types and uses, not companies. Which ad or analytics provider is in the app is a question for the privacy policy, and often answered there only in the plural.
The difference between your tier and someone else's. One label covers the whole app. Apple asks developers to report across all versions — free and paid, regional, adult and child — and Google says its section covers every version on Play, with practices varying by use, region or age. The free tier you are about to try and the top subscription you may end up on share one panel.
Anything resembling an audit. Play can show that an app has had an independent security review, which sounds like verification and is not: Google says the review "does not verify the accuracy and completeness of the developer's Data safety section disclosure". It also notes that an app's permissions list and its Data safety section can differ, because they describe different things.
The label is a statement the developer signed, not a measurement anyone took. That still beats a paragraph of marketing copy — it is a different kind of evidence, and worth weighing as one.
Reading two labels side by side
The comparison is faster than reading either one alone, because you are looking for the difference rather than the content. Open both listings, scroll to the panel, and work down four lines.
What is shared, and why. Two apps can collect the same list and share wildly different amounts of it. Advertising as a declared purpose on a free tier is not a scandal — it is how that tier gets paid for, which we went through on what the ads are told about you — but it belongs in the price you assign the app.
What is optional. Google asks developers to say whether a collection is required or something you can decline. An app that marks several items optional has given you controls before you have spent anything.
The deletion route. Play requires an app that lets you create an account to offer an in-app way to delete that account and its data, plus a web route for requesting deletion. If the panel points you only at a policy page, note it.
The spread between the two panels. If they look identical, the AI girlfriend app privacy label has told you these apps are alike on this axis and the decision sits elsewhere. That is a real result, not a failed check.
The checks the label cannot do for you
A declaration describes intent. Two things on your own phone describe behaviour, both free.
On an iPhone, App Privacy Report records which domains each app contacted over the past seven days and how often it reached for data you granted. It will not tell you what was sent. It will tell you whether a chat app talks to a short list of servers or a long one, which the label does not answer in that form. On Android 12 and newer, the privacy dashboard in Settings shows which apps used the camera, microphone and location in the last day.
Run either for a week and you learn something the panel cannot contain: not whether the developer told the truth, but whether the app behaves the way its description implies.
What the label should and should not decide
It is good at ruling things out and poor at ranking. A listing that declares conversation content shared for advertising has told you something concrete, and you can close it on that basis alone. Two listings that look alike have told you the panel is not the deciding factor, and pretending otherwise dresses up a decision you are making for other reasons.
Where it saves money is the moment before a subscription. An app that gives everyone a deletion route and marks collections optional has handed you for nothing what another may be listing in a paid privacy tier — a trade we picked apart in which privacy settings actually cost money. Read the panel first, then price the tier on features you will use. The app we currently recommend has a free tier, enough to check the deletion path and the optional flags before any card details are involved.
Two minutes, before you install, on a panel most people scroll past. It is not a guarantee and was never meant to be one. It is the only part of the listing somebody had to sign.
Frequently asked questions
Are app store privacy labels verified by Apple or Google?
No. Both schemes are developer declarations. Google states that Data safety information is declared by developers and not verified by Google, and Apple says its App Store privacy information is self-reported by the developer. Treat a label as a claim on the record rather than an audited fact.
Why does an AI girlfriend app privacy label say it collects my messages?
Because it does, and it has to. A companion app sends what you type to a server so a model can produce a reply, which makes message content collected data under both schemes. The useful question is not whether messages are collected but whether they are shared, with whom, and for what declared purpose.
What if the two apps I am comparing have identical privacy labels?
Then the panel has done its job: this is not where they differ. Move the decision to what you can test on a free tier — how the memory behaves, what the paywall interrupts — and hold both apps to the same deletion baseline.
Disclosure. NaughtyEmber may earn a commission if you sign up through a Visit link on this page, at no cost to you. It does not change what we write. How we earn.