Risks and security

Using AI has a price that doesn't always show up on the bill. Every time you hand it a document, an email or a client's data, that information leaves your hands. In this section I explain what I give each AI and what I don't, why, and what precautions I take when I work with client data or with agents that have access to my machines. Privacy, the legal side, what it means to be exposed online and use that gets out of hand all belong here.

We also answer a question few people ask and everyone should: is what I'm doing with AI legal? Personal data and GDPR, what the European AI Act requires of a small business, who owns a generated text or image. Explained in practical terms, with the cases I've actually come across, not in legalese. I'm not a lawyer or a security consultant, and none of this replaces their advice; the law changes fast, so in each article I say when I checked it (most recently in September 2026).

And the quietest risk of all: getting used to AI doing your thinking for you. I tried using it for everything for a week, and I'll tell you where it improved my work and where I stopped thinking for myself. AI is a great tool, but like any powerful tool, it's worth knowing where it can hurt you before it does.