The next early shock was the quiet one: what actually happens to the stuff you type in?
- Your words trained their models. By default, companies used your conversations to improve their models unless you dug into settings and opted out. Then came the leaks, the contractor horror stories, and the regulators knocking. Most big apps now give you toggles to limit training use — but you still have to go and read them. (Lesson 8 goes deep on exactly this.)
- The data they trained on got contested. Authors, news outlets and artists sued over training data — books, articles and images they recognised as their own, scraped without permission. Those cases are still grinding through the courts. It matters to you later: when you build your own stack, you decide what goes in.
- People leaked their own employers. Staff pasted customer data, contracts and internal code into public chats. Still happening today. Almost overnight, "AI policy" became an HR problem.
- Cloning became a fraud tool. Voice and face cloning slid from neat demo to scam — fake "family emergency" calls, bogus CEO wire transfers. Rules tightened around biometrics (your face and voice data) and impersonation.
You don't need the case names memorised. You need the habit baked in: don't paste secrets into rented cloud tools without stopping to think first. That one reflex saves more grief than any setting.
Continue — why the apps feel so censored now.