PowerShift is committed to democratic trade, investment, energy and raw materials policies and accountable to shaping.
Our work lives on careful analysis. Equally important to us the trust of people and organisations, with whom we work together.
That's why we make it transparent, how we use artificial intelligence (AI) and where we consciously draw boundaries. In doing so, we are guided by the requirements of the European AI Act.
What we use AI for
AI applications support us in individual work steps. These include, for example:
- research;
- structuring information and texts,
- linguistic revision of texts;
- translations;
- data analysis and
- organisational tasks.
AI is a tool for us. It does not replace our technical analysis or our political assessment.
What we don't use AI for
We do not use AI to people lead astray.
The following principles apply in particular to our work:
- For PowerShift releases, AI is just a tool.
- We do not attribute invented quotes or statements to real people.
- We do not create photorealistic AI images of real people or events.
- We do not create synthetic sound or video recordings of real people or events, which can be confused with real recordings (‘deepfakes’).
- We give artificially generated content not as documentary evidence.
- We do not fake independent civil society voices.
- We do not invent statements of affected communities.
Responsibility Remains with Humans
For all content that PowerShift publishes, submits or sends; wearing Our employees have the responsibility. This also applies if AI helped create it.
We adopt results from AI applications Not unchecked.
Before publishing, we check in particular:
- facts and figures,
- Sources and quotations, and
- technical and legal assessments.
For this we use the underlying original sources. An AI system itself is not a scientific or journalistic source for us.
We have a binding guideline for the internal handling of AI. All employees are guided by this guideline and are trained in how to use AI.
Labelling and transparency
When we use AI-generated or AI-modified content, We will mark them if such marking is required.
For texts that of people content checked and editorially responsible, we waive usually to a separate AI marking.
Significantly supportive use of AI: For example, for large-scale translations or transcriptions – we do this transparent, if it is objectively necessary or agreed.
Transparency notice from PowerShift e.V.
State of play: August 2026
