EU AI Act: What Applies from Today — and What It Means for Your Family Data

Starting today, 2 August 2026, a new level of the AI Act takes effect in the EU — the world's first comprehensive AI law. From now on, so-called transparency obligations are binding. In short: people should know when they're dealing with artificial intelligence, and artificially generated content should be recognizable as such.

That's good news. And because "AI" now appears on almost every app without anyone explaining what it actually means, we're taking the time to do that here.

**What the Law Requires**

Three things, simplified. First: anyone interacting with an AI system must be able to tell. Second: what an AI produces — texts, images, voices — must be labelled in a machine-readable format so it can be technically verified later. Third: for particularly sensitive applications like emotion recognition or deceptively realistic forgeries of real people, additional, stricter rules apply.

**Which AI is in Famora**

Famora uses AI at precisely the points where family organization is tedious — and nowhere else. You speak an appointment instead of typing it. You photograph a parent letter, and the appointments from it land as a suggestion in your calendar. You forward an appointment confirmation by email instead of typing it out. Then there's the daily overview, recipe suggestions, and the read-aloud function.

The key point: **The AI suggests, you decide.** No appointment, no entry, no change is created without your explicit confirmation. There is no automation in Famora that writes to your family calendar on its own.

**What We've Implemented for This**

Content that an AI has generated — such as the daily overview or a recipe suggestion — is visibly marked in the app. You don't have to guess what comes from a machine.

The read-aloud function produces a synthetic voice. The audio files created in the process now carry a machine-readable label that identifies them as artificially generated. This is the technically most demanding part of the new obligations — and precisely the one you don't see from the outside.

And all AI functions can be switched off. Famora remains fully usable afterwards: calendar, chat, lists, map work without a single AI request.

**What We Deliberately Do NOT Do**

Sometimes the omitted information is the more important one.

Famora does **not analyze emotions**. We do not evaluate biometric features. We do **not replicate the voices of real people** — the read-aloud function uses a standard voice from our service provider, never that of a family member. And we do not use your family data to train AI models.

We care particularly about the voices point. An app that could replicate your child's voice would be technically possible. We decided against it.

**What Goes to Your AI Service Providers — and What Doesn't**

For speech recognition and appointment recognition, we work with specialized providers who act as data processors for us. Only the text content necessary for the respective function is sent to them.

What is **never** transmitted: GPS coordinates, raw location data, complete chat histories, and email addresses of third parties. A forwarded email is deleted immediately after evaluation — only the detected appointment is stored. Which service provider is used for what is fully detailed in our privacy policy.

**Staying Honest**

We deliberately do not write "AI-Act-certified" here. Such certification does not exist in this form, and we don't believe in advertising with seals that no one can verify.

What we can say: we have aligned our AI functions with the transparency requirements that take effect today — and the points above are verifiable. You see the labels in the app. You can switch off the functions. You can read in our privacy policy what goes where.

If something is unclear to you or you find wording that could be better: write to us. That's exactly what transparency is for.