1. Why this page
The Italian law on artificial intelligence (Law 132/2025, Art. 13) requires professionals to use AI only as support and to tell clients, in plain language, which tools they use. The EU AI Act (Art. 50, applicable from 2 August 2026) requires realistic content generated or manipulated with AI to be clearly labelled. This page covers both, simply.
2. In sessions
I only use AI for supporting tasks: research, data analysis, first drafts of text and documents, prototypes of automations and websites. The intellectual work remains mainly mine: I don’t delegate the reasoning, the choices or the responsibility for the result, and I check all content before delivering it. In every proposal I state which tools I will use for that project.
3. In videos
I use generative AI tools to create or edit images, animations, voices and music, and traditional software for editing, colour and audio. The creative choices and the final check are mine.
4. The tools I use
| Tool | What it’s for | Where the data is | Training on your data |
|---|---|---|---|
| [es. ChatGPT, Claude, Gemini · da confermare] | Analysis, research, first drafts | [EU / US with safeguards] | Turned off |
| [es. Higgsfield, Runway, Kling, Veo · da confermare] | Generating and animating video and images | [to check] | [Turned off / not applicable] |
| [es. Midjourney o Flux · da confermare] | Images and graphics | [to check] | [to check] |
| [es. ElevenLabs · da confermare] | AI voiceovers, synthetic voices or with consent only | [to check] | [to check] |
| [es. n8n o Make · da confermare] | Automations in business projects | [EU] | Not applicable |
| Claude (Anthropic) · API | Assistant replying on WhatsApp | United States, under the EU-US Data Privacy Framework and standard contractual clauses | Turned off: data sent through the API is not used for training |
| [es. Premiere Pro o DaVinci Resolve · da confermare] | Editing and finishing | On my computer | Not applicable |
5. How I protect your data
- I only upload to AI tools the material needed for the job.
- I prefer business versions with data processing agreements and training turned off.
- I don’t enter your company’s confidential information without your agreement.
- For automations that process your customers’ personal data we sign an agreement under Art. 28 GDPR.
- I don’t clone voices or recreate the faces of real people without their consent.
6. How I label AI-generated content
Every video containing elements generated or altered with AI comes with a written note and file metadata that say so. No watermarks on the video. Only if it realistically portrays real people, places or events do I add a discreet end card “Made with AI”.
When you publish it, the label is up to you: the “AI-generated content” label offered by Instagram, Facebook, TikTok and YouTube is enough, or a line in the description. The details are in the Terms.
7. The AI assistant on WhatsApp
When you message me on WhatsApp, an AI assistant based on Claude, Anthropic’s model, replies first. I use it directly through Anthropic’s API: messages are processed in the United States, with the safeguards of the EU-US Data Privacy Framework and the standard contractual clauses, and Anthropic does not use data sent through the API to train its models. The assistant tells you it’s an AI in its first message, as the AI Act requires (Art. 50). It only answers with information I have written and approved (services, prices, timing, how things work), it doesn’t sign contracts and it doesn’t ask for payment details in the chat: to book or order, it sends you the link to the right page of the site. You can ask to talk to me whenever you like.
8. The limits of AI
AI can make mistakes or produce unexpected results. That’s why all content goes through my checks, videos include revisions and, in sessions, I verify data and sources before giving you advice. For questions about this page, write to simone@sgl.studio.