Decisions about how AI systems are built are among the most consequential of our time. What values do they encode, whose concerns do they prioritise, what safeguards do they implement and which ones do they overlook? The people most affected by these systems are rarely present when those decisions are made, and typically consulted only once products are already built. As a result, power imbalances are reinforced rather than challenged. And AI developers, deprived of meaningful external input, produce systems that reflect a narrow set of perspectives.
The challenge ECNL set out to address was structural: there was no practical guidance on how to involve affected people in the design, development and deployment of AI systems in a way that produces real change. ECNL was determined to fill that gap.
The Framework for Meaningful Engagement (FME) was developed from 2022 to 2025. It is a practical tool co-created with SocietyInside, and over 300 individuals and groups from civil society, industry and the public sector across the globe. The approach is built around three conditions for meaningful engagement:
- There needs to be a shared purpose that goes beyond the interests of the AI developer and reflects the needs of affected communities and the wider public.
- The process should be inclusive and fair, bringing in a diverse range of stakeholders early enough that they can genuinely influence decisions.
- Engagement should lead to visible impact, meaning that input is taken seriously, feedback is shared with participants, and contributions clearly shape outcomes.
ECNL did not leave the FME as a theoretical resource, and ran two real-world pilot projects to test the framework in practice, each in a distinct context and sector:
- The first pilot was a partnership with Discord, a leading social media platform. ECNL worked with multiple teams within Discord to test whether the FME could guide the development of algorithmic tools for online safety, with a focus on teens.
- The second pilot partnered with the City of Amsterdam as it developed “scan bikes,” an AI-powered image recognition service for public spaces. With ECNL’s support, Amsterdam designed a public engagement process that brought citizens into product design, leading to concrete system changes and increasing trust in the municipality’s willingness to act on feedback.
Both pilots confirmed the central insight that drives the FME: involving stakeholders early, from the stage when ideas are first being developed, rather than after a system is already built increases the chance that their core values are built directly into AI systems. Lessons from the pilots fed directly into the FME 2.0, published in November 2025.
ECNL is building an ecosystem of rights-based AI development: we actively connect AI developers and platforms with civil society organisations ready to engage meaningfully. Going forward, ECNL might conduct future pilots, focused on advanced AI systems such as agentic AI.
The difference this work made:
- Established the first practical, rights-based framework for meaningful engagement in AI development, providing structured guidance for how developers, public institutions and civil society can involve affected communities throughout the AI lifecycle.
- Validated through two real-world pilots that meaningful engagement is possible in both corporate and public-sector AI development, and can lead to concrete changes in AI design.
- Contributed to shifting expectations in the AI field towards more structured and earlier involvement of affected communities.
Image: Hanna Barakat & Archival Images of AI + AIxDESIGN / https://betterimagesofai.org