AI-Karma

Your space · local data

Your assessments

Every assessment stays in this browser only — export it as a JSON file to keep it elsewhere. You can assess several companies or subsidiaries under distinct names and compare them.

New assessmentFree · 15 minutes · no account

Assessment made on the old address (ai-karma.vercel.app) or in another browser? Retrieve the file at ai-karma.vercel.app/en/recuperer, then import it here.

T · Technology

Safety, traceability and explainability of the digital and AI systems in use.

C · Compliance

Alignment with the FADP and, where your activities touch the EU, with the GDPR and the AI Act: records, data subject rights, impact assessments.

E · Environment

Environmental footprint of digital infrastructure and usage.

S · Society

Human impact: fairness, accessibility, transparency and effects on employment.

Your assessments

Loading…

Your compliance tools — free

Does the AI Act apply to you?

For each use of AI in your company, determine whether the European AI Act applies, whether the practice is prohibited or high-risk, and which transparency notices are due — with up-to-date deadlines.

Classify a use case

AI system register

The list of all your AI systems — the one any auditor will ask for, under Art. 26 of the AI Act or under ISO/IEC 42001 (AI management system). It fills itself from your classifications and exports in one click.

Open the register

Transparency pack — Art. 50, applicable since 2 August 2026

The notices that Art. 50 of the AI Act has required since 2 August 2026 for your chatbots, AI-generated content and deepfakes — ready to paste, in French, German, Italian and English.

Generate my notices

Methodology, in brief

Each indicator is scored on a maturity scale from 0 (non-existent) to 4 (systematic). A dimension's score is the average of its 6 indicators; the global score weights the four dimensions according to your sector family (no axis above 40 %). The calculation is a plain weighted sum: every point of the score traces back to a specific answer — no black box. A score of 70/100 opens eligibility for the “Swiss Ethical AI Inside” label (version α, still experimental).