The questions we work on, and the positions we hold.
Artificial intelligence is the defining technology of this period. The questions of how it is built, deployed, and governed will shape decades. This page lays out the issues we work on, the positions we hold, and the frameworks we engage with.
Six questions the AI conversation is converging around.
These are not unique to us. They are the questions that international institutions, nonprofits, regulators, and researchers are working on. We name them here in plain language because half the problem is being able to talk about them clearly.
-
Alignment and safety
AI systems should do what their users and the wider public actually want them to do, and they should fail in ways that are safe, recoverable, and visible. The hardest part of alignment is not technical. It is the politics of who gets to define want.
-
Transparency and explainability
When an AI system makes a consequential decision about a person, the person and their advocates should be able to understand, question, and contest that decision. Black-box systems in high-stakes contexts are a governance failure, not a technical inevitability.
-
AI literacy as a public good
The ability to understand what AI is, what it can do, where it fails, and how to use it well is becoming a precondition for agency in a society where decisions are increasingly mediated by machines. Under Article 4 of the EU AI Act it is also becoming a legal requirement. We treat AI literacy the way the twentieth century treated reading.
-
Who has a seat at the table
Most AI is designed by builders looking down the funnel of their products. The people AI touches, including applicants, claimants, patients, displaced families, and frontline workers, have no seat at the table where the system is built. We work on that asymmetry.
-
Deployment in high-stakes contexts
AI used in asylum, healthcare, education, public services, social protection, and migration is operating in contexts where errors damage human dignity in real and sometimes irreversible ways. The standard for deployment in those contexts cannot be the same as for advertising or entertainment.
-
Concentration of capability
The benefits of AI should not accrue only to the largest companies, the most technical experts, and the wealthiest institutions. They should be understood, questioned, shaped, and used by ordinary people and ordinary organizations.
Our positions on each of these issues.
The roadmap below is the foundation's working position. It is the commitment we are accountable to in the work we do.
-
On alignment and safety
We support the alignment research and the responsible-scaling commitments emerging in the field. In our own scope of work we will not endorse, train, or partner on the deployment of AI systems in high-stakes contexts without explicit human oversight, accountability for outcomes, and the ability to recover from error.
-
On transparency
Every AI tool we review for a partner organization is assessed on the quality of its explainability. Every public guide we publish is written in plain language, in English and Danish, with sources. We hold ourselves to the same standard of transparency we ask of others.
-
On AI literacy
We commit to making AI literacy free at the point of access for our priority audiences: Danish nonprofits, schools, clinics, public-sector teams, small business owners, and mission-driven organizations. We work to translate the Article 4 obligation of the EU AI Act into something organizations can actually do.
-
On who has a seat at the table
Through our public work, our research, and our partnerships, we bring the perspectives of frontline practitioners, nonprofit and community leaders, and the people AI is being deployed on into the rooms where AI systems and policies are designed.
-
On high-stakes deployment
We refuse operational support for the use of AI in the surveillance of people who cannot easily defend themselves, in automated decisions about asylum or fundamental rights, in weapons systems, and in the manipulation of public information. These limits are written into our bylaws and require a high bar of amendment.
-
On concentration
We commit to publishing our research, our tool reviews, and our training materials as open public goods, wherever legally and ethically possible. We work to keep AI literacy from becoming another fault line between those with access and those without.
Aligned with the major responsible-AI frameworks. Specific in our contribution.
The work of the AI Literacy Foundation is part of a broader conversation. We engage with the major international frameworks and regulations shaping responsible AI: the EU AI Act, the General Data Protection Regulation, the OECD AI Principles, the UNESCO Recommendation on the Ethics of AI, and the Council of Europe Convention on Artificial Intelligence.
Our contribution is to bring these commitments into a specific context: Danish civil society, the public sector, small and medium-sized organizations, and the communities most affected by AI deployment. We translate broad principles into practical, usable obligations for organizations and people who do not have direct access to the rooms where standards are written.
"We do not aim to replicate the work of large institutions. We aim to translate it into something usable for those who are downstream of it."
What this looks like in practice.
The roadmap above becomes work. The work is what we do.