20/05/2026
The European Commission has told researchers to document their AI use. The AIR framework can be used to do it.
The EU's updated guidelines on AI in research include a clear instruction for researchers: document your AI use transparently, in the methods section, with enough detail to show how it shaped the research process.
It’s the right call. It’s also easier said than done.
What counts as substantial AI use? How do you describe generative AI involvement in data analysis differently from writing? How do you communicate the level of human oversight that was maintained?
Without a shared vocabulary, transparency statements tend to collapse into a single vague sentence.
The AIR Framework maps the research process across seven stages and five levels of AI involvement, with risk profiles and verification guidance at each point. It gives researchers a structured way to produce the kind of disclosure the EU guidelines are asking for.
I've written a post about how AIR relates to recommendation 2 of the new ERA living guidelines on the responsible use of generative AI in research (link in comments). The AIR Framework itself is free to download from Figshare and published under CC BY 4.0.
The infographic image was created using Gemini Nano Banana 2, then adapted in Photoshop. It's public domain so anyone can reuse it. The QR code leads to the ERA guidelines.
