Personal Privacy Check

See which AI services you actually use in private life, set their privacy settings on purpose instead of by default, and write down where your line runs, for yourself and for the people you help.

Official guidance: IMY (the Swedish data protection authority) on AI and personal data, and the EDPB's Opinion 28/2024 on AI models and Guidelines 03/2026 on web scraping for generative AI. The checklist and the do-not-share line are practice, not research.

Use it when:

• You use one or more AI assistants privately and have never looked at their settings.
• You help a parent, a partner or a teenager set up AI services and want a simple routine.
• A service you use has changed its terms, or added AI features to an app you already had.

Skip it when:

• Your question is about AI use at work: use 10.3 Rules That Apply to My Role and 10.2 What Can Go Wrong.
• You are worried about scam calls or fake videos: use 10.5 Scam & Deepfake Defense.

A chat window feels like a private notebook, but it is a service that processes what you type under its own terms. In the EU that processing falls under ordinary data protection law, and regulators have published guidance on how it applies to AI (IMY; EDPB, 2024). The EDPB has also issued guidelines on collecting web content to train generative models (EDPB, 2026), a reminder that what you post in public can end up in training data too.

The law sets a floor; your settings decide the rest. Most services let you choose whether your chats help train models, how long history is kept and which other data, such as mail, photos or files, the service may reach. No research tells you which choices to make, so this page is practice, not research. It works by turning a vague unease into three concrete acts: list the services, set the settings, and write a line you will hold when convenience pulls the other way. It never asks you to record what you have shared, only what you will and will not share from now on.

Include the obvious assistants and the less obvious AI features in other apps: photo apps, keyboards, mail, social platforms. Describe what each can reach in kinds of data, never in content.

Strong answer: "Photo app, free account, sorting holiday pictures. Can reach: the whole camera roll and its location data." Not "I think it's fine".

Illustrative example: the family and the services are made up.

A parent in Finland fills in the page one evening.

Step 1: three services: a paid chat assistant (can reach his chats and uploaded files), an AI feature in his photo app (the whole family album), and a homework helper his daughter uses (unknown).

Step 2: on the chat assistant he turns off training on his chats, sets history to delete after three months and turns on two-factor sign-in. In the photo app he revokes cloud analysis of the album. The homework helper has no data policy he can find, so it is replaced with one that has.

Step 3: never: ID numbers and card details. Only anonymized: money questions. About family: his daughter's name, school and essays stay out; the homework helper gets the math problem only. He agrees that line with her.

Step 4: re-check in six months, or when any of the three changes its terms. He offers to do Step 2 with his mother on her tablet.