
Your first app on Google Play: what to do before publishing
1 November 2026 • Mobile apps

The model is the easy part. Check what has to work around it before AI reaches real users.
Connecting a language model to an app takes a few hours today. Everything around it is harder: what a user sees when the model is wrong, which data leaves your server and what to do with a problem report. Below is the list we went through for our journaling app.
A person using the app should know when they are reading text from a model. That is a matter of trust and of law. The European Commission states that the AI Act’s transparency rules apply from August 2026. Deferrals concern high-risk systems. Certain providers of systems generating content, on the market before 2 August 2026, have until 2 December 2026 to comply with Article 50(2). Dates have changed, so check them on the official page.
In our app we say so when the account is created, in the reflection statuses and in the report menu on every screen with such content.
If you publish on Google Play, this is mandatory. Google Play’s rules require apps with AI-generated content to let people report offensive content without leaving the app.
A report has to go somewhere. Decide who reads it, how quickly and what happens next. A button that sends a message nowhere meets the requirement only on paper.
A model sometimes declines to answer, and a service can be unavailable. Both states need screens of their own:
For each feature, write down what you send, to whom and how long the provider keeps it. Send only what the model needs. These decisions go into the privacy policy and the store’s data safety form, so prepare them early.
Export and account deletion are best designed from the start. In our app, anyone can download their data as a JSON file and delete the account. Reflections can also be exported to PDF: from the last 7 days in the free version, from any range in Premium.
This list comes from building apps and is not legal advice. Talk to a lawyer if AI touches health, finance, employment or children’s data. The same applies if you are unsure how the AI Act classifies your system.
How content is marked and reported, what happens when the model declines or fails, what data you send, and how export and deletion work. We can go through this list with you for your app.
This article was written with the help of AI. We also used AI to compare its facts with the listed sources.
A few sentences are enough. We usually reply within 24–48 hours.
Contact us