AI Ethics for Researchers

AI ethics for researchers is the discipline of doing AI research in a way that respects participants, society, the field, and the future — and proving that the work was done that way. It sits at the intersection of classical research ethics (Belmont Report, Common Rule, Declaration of Helsinki, CIOMS, IRB process), AI-specific concerns (dual-use research of concern, frontier capability research, agent ethics, synthetic-subject research), participant ethics (informed consent, vulnerable populations, fair recruitment, annotator ethics), data ethics (collection, protection, de-identification, sharing, training-data ethics), publication and reproducibility (COPE / ICMJE norms, NeurIPS reproducibility checklist, AI-assisted research disclosure), the IRB / industry ethics-board landscape, and the funding / industry-tie / conflict-of-interest layer that decides whether research preserves its independence.

This track is written for the practitioners doing this work day to day: AI / ML researchers in academia and industry, frontier-lab safety and capability researchers, applied scientists, PhD students, postdocs, IRB members and chairs, ethics-board members, ethics-clinic consultants, research integrity officers, and the cross-functional partners (legal, privacy, security, policy) who interlock with research. Every topic explains the underlying discipline (drawing on Belmont, Helsinki, CIOMS, the Common Rule, the canonical research-ethics literature, COPE / ICMJE / ACM / IEEE codes, FAccT / AIES / NeurIPS norms, ISO / IEEE 7000 / OECD / UNESCO AI ethics standards, and the lived experience of practitioners who have stood programmes up), the practical methodology that operationalises it, the artefacts and rituals that make it stick, and the failure modes where research ethics work quietly fails to protect anyone.