The UN System Chief Executives Board endorsed ten principles for ethical use across the UN system. They address proportionality and do no harm, safety and security, fairness, sustainability, privacy, human autonomy and oversight, transparency, explainability, responsibility, accountability, inclusion and participation. The principles apply across the AI lifecycle and to UN entities using or developing AI.
What this means for
Reading asOptional: it just picks which take you see first.
Just curious / new to AI
It is the UN system’s principal internal normative framework for responsible AI.
Pick a role above and this becomes the reading for that desk: what it means, what to watch, and a line you can use.
Human Rights
- So what
- Rights risks now have a system-wide internal hook: these principles require privacy, human autonomy and oversight, accountability, inclusion and participation, and do no harm across the full AI lifecycle.
- Watch for
- Track how UN entities turn the principles into their own AI policies, procurement rules, and review processes when they use or develop AI.
- Use it
- Cite CEB/2022/2/Add.1 and test any AI proposal against the listed principles, especially proportionality, privacy, transparency, explainability, responsibility, and accountability.
Data & Digital
- So what
- Ten endorsed, system-wide principles now set the internal baseline for any UN AI you use or build, across the full lifecycle.
- Watch for
- Track how your entity turns these principles into local review steps, procurement gates, and deployment checks.
- Use it
- Cite CEB/2022/2/Add.1 in concept notes, model reviews, and governance checklists; map proposals against all ten principles before approval.
Builders
- So what
- Use this as the UN system baseline for AI you build or deploy: ten principles apply across the lifecycle to both use and development.
- Watch for
- Track how your entity turns these principles into review steps; this sets the system-level floor, not the checklist.
- Use it
- Cite CEB/2022/2/Add.1 in concept notes and map your project to the ten principles before review: safety, privacy, oversight, transparency, accountability, inclusion, and more.
Peace & Security
No Peace & Security reading was written for this one: it was judged to sit outside that desk. It is the UN system’s principal internal normative framework for responsible AI.
Development & Policy
No Development & Policy reading was written for this one: it was judged to sit outside that desk. It is the UN system’s principal internal normative framework for responsible AI.
Front Office
No Front Office reading was written for this one: it was judged to sit outside that desk. It is the UN system’s principal internal normative framework for responsible AI.
OPGA
No OPGA reading was written for this one: it was judged to sit outside that desk. It is the UN system’s principal internal normative framework for responsible AI.
Permanent Missions
No Permanent Missions reading was written for this one: it was judged to sit outside that desk. It is the UN system’s principal internal normative framework for responsible AI.
How other roles read this
Human Rights
- So what
- Rights risks now have a system-wide internal hook: these principles require privacy, human autonomy and oversight, accountability, inclusion and participation, and do no harm across the full AI lifecycle.
- Watch for
- Track how UN entities turn the principles into their own AI policies, procurement rules, and review processes when they use or develop AI.
- Use it
- Cite CEB/2022/2/Add.1 and test any AI proposal against the listed principles, especially proportionality, privacy, transparency, explainability, responsibility, and accountability.
Data & Digital
- So what
- Ten endorsed, system-wide principles now set the internal baseline for any UN AI you use or build, across the full lifecycle.
- Watch for
- Track how your entity turns these principles into local review steps, procurement gates, and deployment checks.
- Use it
- Cite CEB/2022/2/Add.1 in concept notes, model reviews, and governance checklists; map proposals against all ten principles before approval.
Builders
- So what
- Use this as the UN system baseline for AI you build or deploy: ten principles apply across the lifecycle to both use and development.
- Watch for
- Track how your entity turns these principles into review steps; this sets the system-level floor, not the checklist.
- Use it
- Cite CEB/2022/2/Add.1 in concept notes and map your project to the ten principles before review: safety, privacy, oversight, transparency, accountability, inclusion, and more.