Sweden’s Finansinspektionen published guidance on 15 October explaining how existing rules apply when financial firms use AI for ESG data. The first output of its Innovationskompassen programme addresses practical questions raised by the industry, with particular attention to credit institutions’ risk management and lending.[1]
AI use
| Task | Control focus |
|---|---|
| Extracting or structuring data | Traceable sources and verifiable resultsSource 2 |
| Estimating or filling gaps | Methods, assumptions and documentationSource 2 |
| Analysing or informing decisions | Governance, monitoring and reviewSource 2 |
FI says banks should understand external suppliers’ methods and limitations and be able to challenge them. A supplier’s quality assurance normally cannot replace the bank’s own documented assessment.[2]
The scope matters: credit-decision answers exclude natural persons, and the guidance does not resolve SFDR-specific questions. Different uses of the same dataset can require different evidence of quality.[2][3]
Data vendors consequently need to explain provenance and assumptions in a form their bank customers can assess.[2]
Sources & further reading
- AI and ESG guidance launch Finansinspektionen · 15 Oct 2026 · Accessed 21 Oct 2026
- AI and ESG data questions and answers Finansinspektionen · 15 Oct 2026 · Accessed 21 Oct 2026
- Guidance scope Finansinspektionen · 15 Oct 2026 · Accessed 21 Oct 2026