Upturn

Upturn is a Washington, D.C.-based nonprofit research and advocacy organization that advances equity and justice in the design, governance, and use of technology, with a particular focus on how AI systems and algorithms cause concrete harm to marginalized communities in criminal justice, policing, employment, housing, credit, and public benefits. Founded in 2011 by computer scientist Harlan Yu and lawyer David Robinson as a consulting firm, and restructured as a 501(c)(3) in 2017, Upturn's approach is defined by close partnership with civil rights advocates, organizers, and litigators — providing research infrastructure that movement organizations need to contest algorithmic systems in courts and legislative chambers. As Harlan Yu has framed it: "Technology has seeped into the decision-making processes that really matter for people's material conditions."

Founding and structure

Harlan Yu (a computer scientist) and David Robinson (a lawyer) launched Upturn in 2011 to address the growing intersection of technology deployment and civil rights violations. The organization converted to nonprofit status in 2017 and has grown to twelve full-time staff, operating primarily from Washington, D.C. while remaining a largely remote organization. In 2025, Upturn announced an executive director transition: after nearly fifteen years at the helm, Harlan Yu stepped down in June 2026, with Tina M. Park named Executive Director. The board includes Steven Renderos of Media Justice as president — a signal of Upturn's deliberate positioning within the racial-justice civil society ecosystem rather than a purely technical research bracket.

Predictive policing and criminal courts

Upturn's Stuck in a Pattern (August 2016), authored by Logan Koepke and David Robinson, was among the earliest systematic investigations of predictive policing's civil rights implications. Surveying the fifty largest U.S. police departments, the report found that at least twenty had adopted predictive policing systems, often shielded from public scrutiny, and that the systems may not make communities safer while intensifying enforcement in communities already heavily policed. A coalition of seventeen civil rights organizations signed a statement citing the report's findings on systemic bias and lack of transparency. Upturn's 2019 report Civil Rights and Pretrial Risk Assessment Instruments addressed the parallel deployment of algorithmic risk scores in bail and pretrial detention decisions, explaining why civil rights advocates opposed these tools and outlining strategies to limit their impact on incarceration rates. These two bodies of work established Upturn as the primary U.S. research organization interrogating algorithms at the criminal courts nexus — a focus that distinguishes its portfolio from the Algorithmic Justice League, which concentrates on AI bias and facial recognition rather than the criminal-justice deployment layer.

Surveillance and phone forensics

Mass Extraction: The Widespread Power of US Law Enforcement to Search Mobile Phones (October 2020) drew on 110 public records requests to law enforcement agencies across all fifty states and documented more than 2,000 agencies that had purchased mobile device forensic tools (MDFTs). The report showed that law enforcement can extract a complete copy of a phone's data — emails, texts, photos, location history, app data — and that this capacity has been exercised for minor infractions with minimal oversight or warrant requirements. Upturn framed the findings explicitly in terms of racialized police power: MDFTs entrench and extend the same enforcement patterns documented in the predictive policing research. The report drew on a methodological precedent set by The Illusion of Accuracy (December 2017), which examined the reliability implications of allowing police officers unrestricted review of body camera footage before filing reports.

Employment, housing, and public benefits

Beyond criminal justice, Upturn has investigated algorithmic discrimination in hiring — Help Wanted (December 2018) examined predictive hiring tools used by major employers; Essential Work (July 2021) scrutinized fifteen online hourly-employment application processes — and in housing, where How to Seal Eviction Records (July 2022) addressed the downstream algorithmic harm of eviction records circulating in tenant-screening systems. Discrimination Through Optimization (April 2019) demonstrated that Facebook's ad delivery algorithms could distribute job and housing ads along racial and gender lines independent of advertiser targeting choices, contributing directly to civil rights litigation. Upturn has filed amicus briefs in cases ranging from pretrial risk assessment (Philadelphia Community Bail Fund v. Arraignment Court, 2020) to the scope of the Computer Fraud and Abuse Act (Van Buren v. United States, 2020) and cellphone search warrants (Connecticut v. Smith, 2022).

Civil rights partnerships

Upturn's operational model is explicitly research-for-movement: it describes its work as partnering with "local and national advocates, organizers, litigators, academics, and policymakers" and supporting movements building toward a just economic system. Collaborators have noted the organization's willingness to be "of use to those most impacted" rather than positioning itself as a top-down expert authority. In practice this has meant submitting testimony in local police surveillance debates, filing regulatory comments on tenant screening and hiring algorithms, and embedding research findings inside civil rights coalition briefs coordinated through the ACLU and the Leadership Conference on Civil and Human Rights.