Cover: NASA, June 24, 2020; the day CDR was founded.
Speed does not bring safety or accountability.”
Nesibe Kırış Can Program ConsultantThe same tension plays out in organizations with no written policy at all. When a staff member opens a tool to begin research, take meeting notes, draft a text or translate, the organization may not yet have defined its AI policy, who is responsible for what, or its data boundaries. But the moment use begins, the organization has in effect made a decision. Which information can be shared, which output can be trusted, which work must go through human review, and who is accountable when something goes wrong: none of this leaves our responsibility, even without a written rule.
For rights-based organizations, this concern runs even deeper. Beyond using AI themselves, these organizations work with communities affected by surveillance, discrimination, data concentration and digital repression. The issue is therefore not only organizational efficiency; it is a responsibility of trust, care and accountability toward stakeholders.
With this report, we propose a workable path between rejecting technology and adopting it without question.
Drawing on the experiences of the 45 organizations that took part in our survey, we define minimum safeguards to make AI use safer, more accountable and rights-based. Our aim is not a flawless compliance system; it is for every organization to make its current use visible and take the next step from where it stands.
The report sets out to answer four questions.
- For which tasks, and to what extent, do civil society organizations use AI?
- What does this use cost, who pays for it, and is it sustainable?
- What rules and what institutional capacity does responsible use require?
- How can organizations keep their use of AI from harming the people they serve?
Safeguards are not keeping pace with use.
In the 2026 sample, AI use is no longer the exception. 96% of the organizations in our survey said they use AI in at least one area. The key question is therefore no longer whether AI is used, but where it is used and who decides on that use.
- Beginner
- Intermediate
- Advanced
- Expert
Training mostly remains short and informal. 10 of the 45 organizations (22%) said they had received no training or did not know.
The details are in the report.
Use cases, cost and budget, rules and institutional capacity, rights impacts and minimum safeguards are covered in the full report.
This report was prepared by CDR as part of the project. The project is supported by the Digital Spark grant programme under the Digital Democracy Initiative (DDI), within the Regional Support Mechanism EECA implemented by Metamorphosis Foundation and TechSoup in partnership with CIVICUS.
CDR is solely responsible for the content of this report. The views and assessments expressed in it do not necessarily reflect those of CIVICUS, Metamorphosis Foundation, TechSoup or the programme's funders.
