The classic trap
Recital 142 is not binding, but it shapes how surveillance authorities (EU AI Office, future Luxembourg AI authority) interpret applications for public funding or regulatory sandboxes under article 57. In practice, many Luxembourg companies apply to Horizon Europe, Luxinnovation or FNR calls without documenting the social and environmental impact of their AI solution, and lose out to projects that built an interdisciplinary approach from day one (accessibility, non-discrimination, digital rights). CNPD remains competent for the personal data aspects of these projects.
What this recital concretely changes for your funding files
- Public call selection criteria increasingly include an AI for Good grid: WCAG/EN 301 549 accessibility, reduction of socio-economic bias, model carbon footprint.
- Interdisciplinary cooperation becomes an eligibility criterion: a project committee without an accessibility expert or fundamental rights lawyer is a red flag.
- Surveillance authorities can rely on this recital to favourably orient decisions on regulatory sandboxes (article 57) or SME exemptions (article 62).
- Horizon Europe and FNR reviewers already use this interpretive grid to score the societal impact of AI projects.
The practical test before submitting a file
Ask yourself three questions: (1) can I name three concrete social or environmental beneficiaries of my AI solution, measurable with KPIs? (2) does my project team include an accessibility expert, a fundamental rights lawyer and an academic? (3) have I documented the carbon footprint of training and inference of my model? If you answer no to any of these, your file will be set aside in favour of a better-prepared competitor.
How Luxgap automates this risk
Our Luxgap Impact Dossier Builder turns an AI project into a complete and defensible funding file, by automatically cross-referencing your technical architecture with the evaluation grids of Horizon Europe, FNR, Luxinnovation and the criteria of recital 142 of the AI Act. The tool queries your Git repository, your MLOps pipelines (MLflow, Weights & Biases, Azure ML) and your product specifications to generate the social and environmental impact narrative, without the project owner having to write a single page.
- Automatically detects the AI components of your project via scan of the code repository and model registries, and proposes an impact map aligned with the objectives of recital 142.
- Generates the societal impact section of the funding file with measurable KPIs (WCAG 2.2 accessibility, inequality reduction, CO2 footprint per inference using the Green Algorithms methodology).
- Identifies missing expertise in the project team (accessibility, fundamental rights, environment) and suggests real Luxembourg and European academic partners via ORCID and CORDIS APIs.
- Calculates the carbon footprint of training and inference from AWS, Azure or OVHcloud GPU logs, and produces a report defensible before reviewers.
- Produces a timestamped PDF dossier, pre-filled in Horizon Europe, FNR Core and Luxinnovation Fit4Innovation formats, ready to submit.
- Sends real-time alerts on new calls for proposals compatible with your solution via monitoring of the Funding & Tenders and guichet.lu portals.
Available as a complement to a Luxgap DPO or CISO mandate or as a dedicated SaaS module depending on your scope. Request a tailored quote and our teams will prepare a demonstration on your real AI project, with a free 48h white audit to assess your eligibility score for public funding before any engagement.
