Where your data goes with consumer AI
What happens to a document dropped into an AI tool hosted outside Europe, what the terms of use actually say, and the cases where it becomes a real problem.
I build the sovereign AI I talk about.
Your teams already use generative AI, often on services hosted outside Europe, and the AI Act is coming into force in stages. This talk gives your leadership team the bearings to make the call: which data to trust to which models, which obligations are coming, and how to keep control without giving up performance.
Check my availabilityA clear thread, from the threat to the decision, with no scaremongering and no needless legal jargon.
What happens to a document dropped into an AI tool hosted outside Europe, what the terms of use actually say, and the cases where it becomes a real problem.
The timeline for entry into force, the risk levels, and what the regulation asks of a company that uses AI, without reading it for you but giving you the right questions.
How generative AI combines with your existing personal-data obligations, and the reflexes to install in the teams already using it.
European models, hosting in France, on-premise deployment: what is mature, what it costs, and where the trade-off between sovereignty and performance sits today.
Leave with a simple decision grid to sort your use cases, choose where to host what, and bring your teams along rather than banning AI.
I do not comment on digital sovereignty from the outside: I build it. I run a sovereign cloud in France and a meeting-summary AI deployable at the client's site, on sensitive data. So I know first-hand what sovereignty costs, where it keeps its promises and where you need to be clear-eyed; the talk draws on that practice rather than on a literature review.
It is a talk that helps a company use artificial intelligence while keeping control of its data: where it is hosted, which models access it and which rules apply. It connects technical stakes, sovereignty and compliance to inform a leadership decision.
If your teams use AI, very likely yes. The European AI Act applies in stages depending on the risk level of the use case. The talk gives the timeline and the categories to place your own cases, without replacing legal advice.
No, it is not binary. The point is to sort your use cases: what can go through a consumer service and what must stay on sovereign infrastructure. The talk gives a grid to decide use case by use case rather than banning everything.
Executives, leadership teams, CIOs, CISOs, data protection officers, as well as professional clubs and associations. It is accessible without being an engineer, and goes far enough to speak to the IT teams in the room.
Yes. I am based in La Roche-sur-Yon, in Vendée, and I work mainly in the Pays de la Loire and the Grand Ouest, but I travel across France, and in English if your audience needs it.
A short call is enough to nail down the topic, format and date. Tell me your context and your data constraints, and I'll propose the right talk.
Check my availability