Regulatory Consulting at Data & Society Research Institute

Overview

Data & Society is an independent research institute studying the social implications of data, automation, and AI, founded on the conviction that debates about these systems cannot happen meaningfully without an empirical evidence base that challenges both the hype and the fear. Led by Executive Director Janet Haven, it produces original research grounded in the realities of the people most likely to be harmed by how AI is built and deployed, and it has become one of the most cited sources in the algorithmic-accountability field. Its flagship program, the Algorithmic Impact Methods Lab (AIMLab), develops methodologies for conducting empirical, participatory algorithmic impact assessments to support real AI governance — not paper compliance. A second major research stream, ‘AI on the Ground,’ produces empirical work that anchors AI policy and public accountability in lived experience, spanning both the people who build and procure AI systems and the people who use and are affected by them. Data & Society’s influence on federal policy is concrete: the NTIA’s AI Accountability Policy Report extensively cited the institute’s comment and reflected many of its positions. Funded by foundations including MacArthur, Mellon, Annie E. Casey, and Knight, it operates as a nonprofit whose independence lets it interrogate industry narratives directly.

AI-Focused Regulatory Work

Data & Society’s distinctive contribution is turning ‘algorithmic accountability’ from a slogan into a testable empirical question — and then reporting honestly when the answer is uncomfortable. Its AIMLab researchers took on one of the field’s most consequential natural experiments: New York City’s pioneering AI-in-hiring bias-audit law. Working with Cornell University and the UK’s Ada Lovelace Institute, they produced a pair of FAccT papers — ‘Auditing Work’ (which won Best Paper in its category) and ‘Null Compliance’ — showing how the landmark law ultimately falls short of its own goals. Researcher Briana Vecchione has separately built a taxonomy of AI audit practices, concluding that the conditions necessary for an effective audit are rarely met in practice. That willingness to test whether accountability mechanisms actually work is precisely what makes the institute valuable to policymakers designing the next generation of AI rules. Its portfolio ranges widely and pointedly: ‘Red-Teaming in the Public Interest’ (2025), research on the surging environmental costs of AI, work on how people turn to general-purpose chatbots for emotional support and mental-health care in the absence of affordable alternatives, and its foundational ‘Algorithmic Accountability: A Primer,’ originally prepared for the Progressive Congressional Caucus’s Tech Algorithm Briefing. Its researchers testify and engage directly — including before the House Oversight Committee — and Executive Director Janet Haven sits on Partnership on AI’s AI & Philanthropy Steering Committee alongside peers from the Ada Lovelace Institute, AI Now, and major foundations. For anyone mapping the accountability wing of AI policy, Data & Society is the shop that supplies the evidence others cite.

Notable AI Policy Work

• Runs the Algorithmic Impact Methods Lab (AIMLab), developing methodologies for empirical, participatory algorithmic impact assessments to support real-world AI governance.
• Produced the FAccT papers ‘Auditing Work’ (Best Paper in category) and ‘Null Compliance’ with Cornell and the Ada Lovelace Institute, showing how NYC’s landmark AI-in-hiring bias-audit law falls short of its goals.
• Published ‘Red-Teaming in the Public Interest’ (2025) and a taxonomy of AI audit practices identifying the rarely-met conditions for an effective algorithmic audit.
• Shaped federal policy: NTIA’s AI Accountability Policy Report extensively cited Data & Society’s comment and reflected many of its positions on AI accountability.
• Runs ‘AI on the Ground,’ empirical research grounding AI policy in lived experience — including studies of people using general-purpose chatbots for emotional support and mental-health care, and of AI’s surging environmental costs.
• Published ‘Algorithmic Accountability: A Primer,’ originally prepared for the Progressive Congressional Caucus’s Tech Algorithm Briefing, and testified before the House Oversight Committee.

Thought Leadership

1. Data & Society 2025 Annual Report
   Institutional highlights from 2025 — challenging industry narratives, grounding AI in lived experience, and reorienting AI for the public interest.
2. Algorithmic Accountability: A Primer
   The field-defining primer on assigning responsibility for algorithmic harm, covering fairness and bias, opacity and transparency, and the absence of auditing standards.
3. Contributions to the Field: AI Accountability, Infrastructural Struggles, and the Surveillance of Labor
   Summarizes the AIMLab work on NYC’s AI hiring-audit law (‘Auditing Work,’ ‘Null Compliance’) and a taxonomy showing effective AI audits are rare in practice.
4. AI on the Ground (research program)
   Empirical research grounding AI policy in lived experience — from how AI systems are procured and sustained to how people use chatbots for emotional support.
5. Research and Resource Library
   Data & Society’s full library of research on the social implications of data, automation, and AI.

Clients

Independent nonprofit — no commercial clients. Funded by foundations including the John D. and Catherine T. MacArthur Foundation, Andrew W. Mellon Foundation, Annie E. Casey Foundation, and the John S. and James L. Knight Foundation. Research partners include Cornell University and the Ada Lovelace Institute; policy engagement spans NTIA, Congress (House Oversight, Progressive Congressional Caucus), and state legislatures.