Legal
AI Transparency
This page explains where artificial intelligence is used on a-ops.de, how results are reviewed, and how you can identify AI-assisted content.
Last updated: 9 November 2026
Why this page exists
The transparency obligations under Article 50 of the EU AI Act have applied since 2 August 2026. Among other things, they require that people can recognise when they are interacting with an AI system, and that certain AI-generated content is labelled.
A statutory exemption applies to some of the content here because it undergoes editorial review. I disclose it anyway. Anyone writing about AI governance should not fall back on exemptions for their own website.
Where AI is used
| Area | System used | Purpose |
|---|---|---|
| Blog texts | DeepSeek (deepseek-chat) | Drafts and research, always editorially revised |
| Header images | AI image models | Abstract and illustrative post images |
| Newsletter | DeepSeek (deepseek-chat) | Draft of the edition in the editorial area, sent by hand |
Some older header images were generated with other image models. Most carry a signed provenance record (C2PA) inside the image file.
No AI is used by Ask a-ops with its full-text search, the situation helper on the tools page and the ways for agents (For agents): they return passages and articles, no generated answers.
Which providers process data in this context is described in the privacy policy.
How you can identify it
Every post carries one of two labels in its header if AI was involved:
- AI-assisted research, text written editorially Research was AI-assisted, the text is mine.
- AI-assisted, editorially reviewed The draft came from an AI system and was revised and approved on the merits.
Posts without a label were created entirely without AI. When in doubt I label rather than omit. This also applies to older content: where the origin of an older post can no longer be reconstructed with certainty, it is marked as AI-assisted.
Editorial responsibility
No post on this website is published unreviewed. Every text is checked on the merits before publication: facts, figures and sources are verified, reasoning and language are revised. Editorial responsibility rests with Mario Lohe.
Newsletter: I read every edition before it is sent. If it was drafted with AI, the edition says so.
That does not mean errors are impossible. If you spot one, get in touch.
Images
Header images are almost entirely abstract or clearly illustrative. They do not depict real people or real events. Where an image looks photorealistic, the notice appears directly beneath it.
I do not use AI images that depict real people or that could be mistaken as evidence of an actual event.
What I do not do
- No AI-generated quotes, studies or statistics without verification against the original source.
- No AI-generated testimonials, references or case examples presented as genuine.
- No automated publication of posts without human approval.
Questions
Write to [email protected].