Responsible AI Ecosystems in Banking and Insurance

A bank or an insurer cannot govern its AI systems alone. Supervisors, standard-setters, auditors, cloud and model providers, software vendors, insurers, customers and civil society all take part in deciding what responsible AI means and whether it has been achieved.

The EU has called for an ecosystem approach to responsible AI, but what this ecosystem looks like in practice, and what it means for high-trust, heavily regulated sectors such as banking and insurance, is still unclear.

FLAIG’s ecosystem theme treats this set of actors as a responsible AI ecosystem and asks two questions: What does the ecosystem look like? And how should a bank or insurer position itself within it?

  • Main results:
  • 1. Responsible AI ecosystem model. The model describes AI governance on three layers: the societal and ecosystem layer, the organizational layer, and the AI system layer. Between organizations, governance takes two forms. In polycentric governance, independent external actors scrutinise the firm and collaborate with it. In value chain governance, obligations run through the firm’s relationships: upstream with technology vendors, downstream with customers, and through insurers who help alleviate AI risk.
  • 2. AI ecosystem strategy framework for banking and insurance. The framework turns the model into seven decision areas that a bank or insurer must settle, from external scrutiny and vendor governance to customer-facing governance and anticipatory capability. For each area, it describes three positions — minimal compliance, managed participation and active shaping — and four tests for choosing between them. The framework is published as version 1.0 for validation with partners.
  • 3. Piloted horizon-scanning process. Through an expert study on generative AI and its governance, the project piloted a foresight process that banks and insurers can repeat in a lighter form every year.
  • Plural logics, translated into practice
  • - Actors in the ecosystem do not share one idea of what “responsible AI” means. They draw on competing institutional logics: AI ethics principles, managerial rationalism, IT professionalism and regulatory oversight.
  • - External requirements do not apply themselves. They are translated into practice through negotiation inside each organization, which is why formally similar governance arrangements can work very differently in practice.
  • - Responsibility for AI is therefore distributed across the ecosystem rather than located in a single function or organization.

Publications and outreach:

Reports

Minkkinen, M. (2026). Responsible AI Ecosystems in Banking and Insurance: An Ecosystem Model and a Strategy Framework. FLAIG Work Package 2 report. University of Turku.

Scientific publications

Khan, F., & Mäntymäki, M. (2027). Interorganizational AI governance in academic and grey literature. In I3E 2026, LNCS 16912, pp. 326–336. Springer. https://doi.org/10.1007/978-3-032-38014-2_24

Birkstedt, T., Minkkinen, M., & Mäntymäki, M. (2027). Governance by negotiation: How organizations translate AI ethics into practice. In I3E 2026, LNCS 16912, pp. 301–312. Springer. https://doi.org/10.1007/978-3-032-38014-2_22

Minkkinen, M., Jylhä, H., & Mäntymäki, M. (2026). AI governance: A layered institutional perspective. Working paper. (unpublished)

Jylhä, H., Minkkinen, M., & Mäntymäki, M. (2026). Race against or with the machine? From perceived limitations, concerns, and future potential of GenAI to anticipatory governance. ISPIM Conference Proceedings.

Master’s theses

Heikkilä, S. (2026). Responsible AI ecosystems: Insights from the insurance industry. Master’s thesis, Information Systems Science, University of Turku. Supervisor: Matti Minkkinen. https://urn.fi/URN:NBN:fi-fe20261005131429

Huopalainen, J. (2026). Customer expectations of AI in retail banking: A vignette-based qualitative study. Master’s thesis, Information Systems Science, University of Turku. Supervisor: Matti Minkkinen. https://urn.fi/URN:NBN:fi-fe20260911125025

Presentations and talks

3.11.2026, FLAIG Summit: “Responsible AI ecosystems: plural logics in banking and insurance”, Matti Minkkinen (University of Turku).

15.9.2026, I3E 2026 conference, Tampere: “Interorganizational AI governance in academic and grey literature”, Fuad Khan (University of Turku).

15.9.2026, I3E 2026 conference, Tampere: “Governance by negotiation: How organizations translate AI ethics into practice”, Teemu Birkstedt (University of Turku).

Teaching

4.2.2026, “Responsible AI: AI governance, business models, and ecosystems”, guest lecture, Matti Minkkinen (University of Turku).

The research approaches include:

  • Conceptual theory synthesis (institutional logics, systems theory, business ecosystem theory)​
  • Synthesis of academic and grey literature on interorganizational AI governance​
  • Expert interviews in Nordic banks and insurers​
  • Expert interviews in the Finnish insurance sector​
  • Vignette-based interviews with retail banking customers​
  • Expert elicitation for horizon scanning​