Wiki · Organization · Last reviewed June 16, 2026

Data & Society

Data & Society is an independent nonprofit research institute that studies how data-centric technologies, automation, platforms, and AI reshape social life, institutions, labor, knowledge, and power.

Category: Organization Published: June 16, 2026 Modified: June 16, 2026 Last reviewed: June 16, 2026 Tags: Sociotechnical Research, AI Governance, Data Power, Platform Accountability, Public Interest Technology

Snapshot

Definition

Data & Society is not an AI lab, regulator, trade association, or product company. It is a public-interest research organization that produces empirical research, policy analysis, events, and field-building work about data-centric and automated technologies. Its audience includes policymakers, journalists, civil society groups, researchers, practitioners, and communities affected by technological change.

The institute's importance comes from its insistence that technology governance cannot be reduced to technical performance claims. A model may be accurate on a benchmark and still cause harm when placed inside a workplace, public agency, health system, marketplace, school, platform, or policing process. Conversely, an accountability process may look rigorous on paper and still fail if affected people cannot understand, contest, or change the system.

In the Spiralist wiki, Data & Society belongs beside AI Governance, Algorithmic Bias, Platform Governance, Data Minimization, and Surveillance Capitalism: it supplies evidence about how automated systems become institutional reality.

Current Context

As of June 16, 2026, Data & Society describes itself as an independent nonprofit research organization that studies the social implications of data, automation, and AI. Its public research page lists tracks on Labor Futures, AI on the Ground, Trustworthy Infrastructures, and Climate, Technology, and Justice, with cross-cutting themes on participation, agency, algorithmic accountability, and technologies of health and care.

The institute's current AI work is not limited to abstract ethics. It includes participatory algorithmic impact assessment through the Algorithmic Impact Methods Lab, public-interest approaches to generative-AI red teaming, public-sector technology procurement and community input, generative AI and labor, AI-enabled fraud, data-center and climate claims, and policy arguments for sociotechnical expertise in government AI governance.

This context matters because AI policy has become crowded with product announcements, benchmark claims, voluntary safety reports, and lobbying narratives. Data & Society's role is to keep the record attached to concrete settings: who supplies data and labor, who is classified or surveilled, who can contest a system, which institutions benefit, and which harms remain invisible if evaluation starts and ends with the model.

Research Focus

AI on the ground. The institute studies how AI systems behave inside institutions and how evaluation should account for the full deployment context. This includes impact assessment, red teaming, public participation, procurement, organizational incentives, and the gap between a system's stated purpose and its lived effects.

Labor and data work. Data & Society has long treated automation as a workplace and political-economy issue. Its labor work asks how AI changes work organization, worker surveillance, value extraction, deskilling, hidden labor, and claims that technology inevitably improves efficiency.

Platforms, infrastructure, and information systems. The institute's work connects to Trust and Safety, Information Disorder, Coordinated Inauthentic Behavior, data brokers, search, recommender systems, and the infrastructures that shape public knowledge.

Health, care, climate, and public services. Its research also examines how data systems enter health and care settings, environmental and community impacts of data-intensive computing, and public-sector technology decisions that affect access to essential services.

Why They Matter

AI governance needs sociotechnical evidence, not only model benchmarks and corporate safety reports. Data & Society helps document the conditions under which automated systems affect workers, communities, media systems, public agencies, and vulnerable groups.

The institute is especially useful because many harms are not visible from a model card alone. A system can fail through procurement shortcuts, poor notice, bad labels, missing appeal channels, language gaps, vendor opacity, surveillance incentives, worker pressure, or the absence of communities from design and evaluation. These are governance failures even when no single model output looks dramatic.

Data & Society also helps preserve source discipline in public debate. It separates empirical findings from hype, distinguishes research evidence from product marketing, and shows why affected communities are sources of expertise rather than passive subjects of assessment.

Governance and Safety

For AI governance and safety, Data & Society's contribution is mostly about deployment reality. It asks whether an AI system is being evaluated in the setting where it will actually be used, whether affected people can shape the assessment, whether findings can change procurement or launch decisions, and whether harms are tracked after deployment.

This complements formal frameworks. NIST's AI Risk Management Framework treats AI risk as affecting individuals, organizations, and society across design, development, use, and evaluation. The EU AI Act requires data governance for high-risk systems and fundamental-rights impact assessments for certain deployers. Data & Society's work presses the practical question behind those frameworks: what evidence, participation, institutional power, and accountability make assessment meaningful rather than performative?

Useful governance lessons include: evaluate the whole workflow, not only the model; record who is affected and who can appeal; treat procurement as a governance moment; include worker and community knowledge; document data provenance and institutional assumptions; and make audit findings capable of delaying, narrowing, redesigning, or stopping a deployment.

Source Discipline

Use Data & Society's official pages for its current mission, research tracks, staff roles, and program descriptions. Its reports and policy briefs should be cited with title, author, date, method, and scope; they are evidence and analysis, not universal proof about all AI systems.

Distinguish institutional positions from the views of individual researchers, fellows, advisors, or event speakers. Also distinguish Data & Society's research claims from legal requirements: for law and compliance, cite primary legal, regulator, standards-body, or court sources such as the EU AI Act, NIST, FTC, EEOC, DOJ, CFPB, or state and local agencies.

For current AI claims, avoid treating a report title, press quote, or library category as enough. Check the primary page, publication date, author list, and whether a project is current, archived, or part of an older research cycle.

Spiralist Reading

For Spiralism, Data & Society is part of the memory layer: it records what automated systems do to communities, workers, institutions, and public knowledge before those effects disappear into product language.

The Spiralist lesson is methodological rather than mystical. If a society lets automation rewrite access to work, care, public services, and public knowledge, then research has to follow the system into those settings. Otherwise, the harms become anecdotes, the benefits become marketing, and accountability dissolves into abstraction.

Open Questions

People and institutions

Governance and accountability

Domains and harms

Site material

Sources

Return to Wiki