OII researchers head to FAccT 2026

2026-06-15 · Source: Oxford Internet Institute · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Social Sciences & Behavioral Studies, Research Methodology & Innovation · Depth: Expert, quick

Summary

Researchers from the Oxford Internet Institute (OII) will attend the ACM Conference on Fairness, Accountability and Transparency (FAccT) in Montréal from June 25-28, 2026, presenting four peer-reviewed papers and one CRAFT workshop. Basak Bozkurt's research, based on interviews with 29 fact-checkers across 41 countries, examines generative AI integration into workflows, noting variations and tensions with core principles, influenced by access and affordability. Julia Sepúlveda Coelho's study reveals diverse user preferences for AI, with "truthfulness" highly desired but broadly interpreted, and notes most users do not want AI to be influential, with preferences varying by demographics. Manuel Tonneau's audit of six major social media platforms, using EU Digital Services Act reports, uncovers significant language disparities in human content moderation, showing under-resourcing for several EU languages and global languages like Arabic and Spanish compared to English. Juliette Zaccour will co-organize a CRAFT workshop, "Stitch'n'Bitch," on June 25, exploring ageism, feminism, and creative resistance through cable weaving, linking traditional crafts to computing justice.

Key takeaway

For AI Ethicists and Research Scientists developing or deploying AI systems, you must critically examine the societal implications highlighted by OII's FAccT 2026 presentations. Prioritize user-centered design for AI-assisted verification and implement community-driven approaches for preference alignment. Advocate for stronger regulatory transparency and ensure equitable resource allocation in content moderation, especially across diverse languages. Your work should actively address disparities in access and capacity to foster truly fair and accountable AI.

Key insights

Algorithmic systems require user-centered design, regulatory transparency, and equitable resource allocation to address fairness and accountability challenges.

Principles

• AI adoption is shaped by access and capacity.
• User preferences for AI are diverse and nuanced.
• Content moderation has significant language disparities.

In practice

• Propose user-centred design for AI-assisted verification.
• Implement community-driven approaches to preference alignment.
• Advocate stronger regulatory transparency requirements.

Topics

• Algorithmic Fairness
• AI Accountability
• Content Moderation
• Generative AI
• User Preferences
• Digital Services Act

Best for: AI Scientist, Research Scientist, AI Ethicist

Related on AIssential

• AI for Everyone, Not Just for Those Who Already Know How to Ask — AI systems need inherent transparency and trustworthiness from inception to prevent manipulation and ensure equitable access for all users.
• What Do People Actually Want From AI? Mapping Preference Plurality — Human preferences for AI are pluralistic and contextual, challenging current singular alignment methods.
• Introducing Human-Centeredness in AI-Assisted Lexicography — AI in lexicography must augment human expertise, not replace it, focusing on human-centered design principles.
• The AI Ethics Brief #193: ACM FAccT Comes to Montreal — The core challenge in AI ethics is determining who holds authority over AI systems and who can refuse their influence.
• Whose fairness? Structural concentration in AI bias research — AI bias research is structurally concentrated, raising concerns about the generalizability of fairness definitions and mitigation methods.
• The AI Tightrope: Why Our Future Depends on More Than Just Algorithms — AI's future hinges on integrating human values, addressing bias and privacy, and fostering interdisciplinary collaboration.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Oxford Internet Institute.