AI Transparency Statement

Last reviewed July 2026

We help small organisations use AI well, so it matters that we are open about how we use it ourselves. This page explains, in plain language, where AI shows up on this website and in our work, what stays under human control, and what we deliberately choose not to do. It sits alongside our Privacy Policy, which covers the data side in detail — this page is about the AI itself.

Where we use AI

In Short: We use AI as a drafting and productivity tool, always with a human reviewing the result before it reaches you.

  • Website content. Some of the writing on this site is drafted with the assistance of AI tools and then reviewed and edited by a person before publication. The ideas, claims, and final wording are ours.
  • French translation. Most French-language pages were translated from English with AI assistance and then reviewed by a human. Only public website copy is translated this way — never personal data.
  • Newsletter curation. Our newsletter is curated with the help of AI (Mistral), which summarises publicly available news articles. It does not access subscriber data.
  • Internal productivity. We use AI tools in our own work the same way we help clients do — for drafting, research, and analysis — with a person accountable for anything that gets acted on or published.

What we deliberately don't do

In Short: No hidden AI, no automated decisions about people, and none of your data used to train anyone's models.

  • We do not run a chatbot that pretends to be a human. The readiness check on our AI Act readiness page is simple built-in logic that runs in your browser — it is not an AI system and it stores nothing.
  • We do not use AI to make automated decisions about you, to profile visitors, or to score or rank people.
  • We do not use AI to generate images or video. Building or editing visual templates in code is a different thing — but we do not produce AI-generated pictures or footage.
  • We do not send website-visitor data to AI providers, and we do not let vendors train their models on client or visitor data.
  • We do not present AI-generated content as if a person had written every word — hence this statement and the named editorial responsibility below.

Human oversight and editorial responsibility

In Short: A named person is accountable for what we publish.

Aisling McCaffrey holds editorial responsibility for the content on this site, including material drafted with AI assistance and reviewed before publication. AI helps us work faster; it does not get the final say. You can read more about us.

AI in our client work

In Short: We use the best models for the job, inside an architecture that keeps our clients' data sovereign.

When building for clients we may use leading AI models, including non-European ones, where they are genuinely the right tool. What stays sovereign is the architecture around the final product or implementation: EU-hosted infrastructure, open-source tooling, and data that remains under the client's control. Sensitive personal or confidential material is processed only on EU-hosted or local open-weight models, never routed to non-European providers. This is governed by each client agreement; see our services for how we work.

Your data and AI

In Short: The AI we use on this site does not process your personal information.

The AI uses described above operate on our own content and on public information — not on your personal data. Our website uses no cookies and privacy-friendly, EU-hosted analytics (Plausible). For the full detail on what we collect and how we handle it, see our Privacy Policy and Cookie Policy.

Our AI Act transparency commitments

In Short: If we ever add AI that interacts with you or publishes realistic synthetic media, we will tell you clearly, at the point it happens.

The EU AI Act's transparency rules (Article 50) apply from 2 August 2026. Today this site has no AI chatbot and publishes no realistic AI-generated depictions of real people or events, so those point-of-use disclosures are not currently triggered. Should that change — for example, if we introduce an AI assistant or publish realistic synthetic media — we will label it plainly at the moment you encounter it, not buried in a policy. If you want to understand how these rules might apply to your own organisation, our AI Act readiness page walks through it.

Questions

If anything here is unclear, or you would like to know more about how a specific feature uses AI, email us at info@crystallized.lu.

Updates to this statement

We review this statement as our use of AI evolves and as guidance under the AI Act develops. The date at the top reflects the last review.