One letter about how a text was made.
AISM describes how an article was made. It does not rate its quality, truthfulness or the worth of its author.
The public label is a single signal: A — human authored; B — human + AI; C — co-created; D — AI + human; E — AI generated. A question mark means that the declaration is incomplete. It is neither an accusation nor evidence of AI use.
The label is based on the actual process: concept, source research, structure and final wording. Style, a detector or a watermark alone is insufficient. The letter applies to the text; photographs and illustrations have separate origin labels.
This is a pilot scale. Its thresholds need calibration and research. An editorial declaration is not an independent audit, certification or ISO standard. To request a correction, use the form on the policies page and include the entry ID.
Scale and calculation
A: 0–10%; B: over 10–30%; C: over 30–60%; D: over 60–85%; E: over 85–100% weighted involvement of generative AI operations. Concept carries 10%, research 20%, structure 25%, final wording and editing 45%. This is not the percentage of generated words.
AISM originated in the Beata at the royal courts project. Concept and direction: Marcin. Prototype developed with support from OpenAI ChatGPT.
Policies and correction requests ↗