SRI researchers present work at ACM FAccT 2026
SRI researchers will present 13 accepted papers and one workshop session at ACM FAccT 2026, showcasing the Institute’s research in responsible AI, algorithmic governance, alignment, privacy, and the impacts of socio-technical systems. (Image credit: Google DeepMind/Unsplash.)
Researchers from the Schwartz Reisman Institute for Technology and Society (SRI) will have a significant presence at the ACM Conference on Fairness, Accountability, and Transparency (FAccT) 2026, the leading international conference on fairness, accountability, and transparency in socio-technical systems, taking place in Montréal from June 25 to 28.
This year, SRI is represented by 19 researchers across 13 accepted papers and one workshop session. Together, these contributions reflect the depth and breadth of SRI’s research community, with work spanning AI governance, public-sector algorithms, privacy, human rights, data annotation, AI companionship, pluralistic alignment, computer graphics, and the social and political dimensions of responsible AI.
Advancing fairness, accountability, and transparency
Several accepted papers examine how AI systems are documented, governed, and held accountable in public and institutional settings. Syed Ishtiaque Ahmed and Shion Guha investigate Canada’s Federal AI Register, analyzing what the transparency tool reveals about government AI use while also identifying what it omits or obscures about human discretion, uncertainty, and accountability. Parand A. Alamdari, Toryn Klassen and Sheila McIlraith bring formal methods together with large language models, proposing approaches for auditing, runtime monitoring, and intervention when advanced AI systems risk violating safety constraints, rules, norms, or regulations. Noam Kolt contributes to the 2025 AI Agent Index, which documents the technical capabilities, safety features, and transparency practices of deployed agentic AI systems.
Other papers focus on alignment, human judgment, and the values embedded in AI systems. Steve Coyne examines the role of human annotators in reinforcement learning from human feedback, distinguishing between annotation as an extension of designer judgment, evidence about social or moral facts, and a form of democratic authority. Marzyeh Ghassemi contributes work that measures how large language models respond to human rights claims across different identity groups, showing how hedging and non-affirmation can produce unequal treatment in contexts where rights should apply universally. Nicolas Papernot and Gillian Hadfield explore how insights from social, economic, and contractual alignment can improve LLM alignment, arguing for approaches that better reflect the complexity and uncertainty of human values. Travis LaCroix advances this theme through work on pluralistic alignment, framing AI alignment as a governance challenge involving competing principals, objectives, and trade-offs, as well as a second paper on how terms such as “alignment,” “agent,” and “hallucination” can shape AI hype, power, and public understanding.
SRI researchers are also contributing work on privacy, public services, platform power, and the politics of data. Elliot Creager examines how differential privacy affects algorithmic collective action, asking how privacy-preserving machine learning can complicate grassroots efforts by users, workers, or citizens to influence AI systems through the data they share. Erina Moon and Guha analyze algorithmic prioritization in the public sector, showing how tools designed to allocate scarce resources can intensify inequality when deployed under real-world resource constraints. Patrick Yung Kang Lee, Jessica Bo, Ashton Anderson and Anastasia Kuzminykh examine AI companionship, drawing on community discussions, surveys, and interviews to understand how users negotiate agency, emotional connection, platform control, and model instability in relationships with chatbots. Julian Posada contributes two papers: one that examines the racial assumptions embedded in computer graphics research on skin and hair, and another, with Ahmed, that critiques the “ground truth” paradigm in data annotation and argues for pluralistic annotation infrastructures that treat disagreement as meaningful rather than noisy.
SRI researchers will also take part in FAccT’s CRAFT program, which supports critical, community-driven and experimental engagements with computing and AI. Ramaravind Kommiya Mothilal will lead a workshop session alongside Ahmed and Guha unpacking assumptions and ambiguities in Canada’s Algorithmic Impact Assessment.
The Institute congratulates all SRI researchers and collaborators whose work was accepted to FAccT 2026 and looks forward to celebrating their contributions in Montréal. See the full list of FAccT 2026 accepted papers.
Accepted papers
Syed Ishtiaque Ahmed and Shion Guha, with Dipto Das and Christelle Tessono, “Bureaucratic Silences: What the Canadian AI Register Reveals, Omits, and Obscures”
Parand A. Alamdari, Toryn Q. Klassen and Sheila A. McIlraith, “Formal Methods Meet LLMs: Auditing, Monitoring, and Intervention for Compliance of Advanced AI Systems”
Steve Coyne, “Three Models of RLHF Annotation: Extension, Evidence, and Authority”
Elliot Creager, with Rushabh Solanki, Meghana Bhange and Ulrich Aïvodji, “Crowding Out the Noise: Algorithmic Collective Action Under Differential Privacy”
Marzyeh Ghassemi, with Rafiya Javed, Cassandra Parent and collaborators, “Hedging and Non-Affirmation: Quantifying LLM Alignment on Questions of Human Rights”
Shion Guha and Erina Moon, “The Paradox of Prioritization in Public Sector Algorithms”
Noam Kolt, with Leon Staufer, Kevin Feng, Kevin Wei and collaborators, “The 2025 AI Agent Index: Documenting Technical and Safety Features of Deployed Agentic AI Systems”
Travis LaCroix, “Relative Principals, Pluralistic Alignment, and the Structural Value Alignment Problem”
Travis LaCroix, with Fintan Mallory and Sasha Luccioni, “Strategic Polysemy in AI Discourse: A Philosophical Analysis of Language, Hype, and Power”
Patrick Yung Kang Lee, Jessica Bo, Ashton Anderson and Anastasia Kuzminykh, with Zixin Zhao and collaborators, “Large Language Lovers: Lived Experiences of Negotiating Agency and Platform Control in AI Companionship”
Nicolas Papernot and Gillian Hadfield, with Karolina Stanczak, Nicholas Meade, Mehar Bhatia and collaborators, “Societal Alignment Frameworks Can Improve LLM Alignment”
Julian Posada, with Theodore Kim, Alexa Schor and Alka V. Menon, “The Racial Character of Computer Graphics Research”
Julian Posada and Syed Ishtiaque Ahmed, with Sheza Munir, Benjamin Mah, Krisha Kalsi, Shivani Kapania and collaborators, “The Consensus Trap: Dissecting Subjectivity and the ‘Ground Truth’ Illusion in Data Annotation”
CRAFT workshop session
Ramaravind Kommiya Mothilal, Syed Ishtiaque Ahmed and Shion Guha, with Faisal Lalani, Dipto Das and Sharifa Sultana, “Unpacking Assumptions and Ambiguities in Canada’s Algorithmic Impact Assessment,” June 27, 3:30 - 4:30 PM.
About FAccT
The ACM Conference on Fairness, Accountability, and Transparency (FAccT) is a leading interdisciplinary venue for research on the social, technical, legal, ethical, and policy dimensions of computing and AI systems. Bringing together researchers and practitioners from fields including computer science, law, social science, humanities, policy, and design, FAccT focuses on how socio-technical systems are built, evaluated, governed, and experienced in the world. The 2026 conference will be held at Le Centre Sheraton Montréal from June 25 to 28.
Want to learn more?
Browse stories by tag:
- AI
- AI safety
- AI trust
- Computer science
- Computer security
- Copyright
- Cybersecurity
- Data
- Democracy
- Economics
- Education
- Engineering
- Ethics
- GPO-AI
- Governance
- Health
- Human Rights
- In the Media
- Jobs
- LLMs
- Law
- Normativity
- Philosophy
- Political Science
- Privacy
- Privacy Series
- Psychology
- Public Policy
- Recommenders
- Regulation
- Religion
- Reports
- Trust
- Workshops
