- Published on
An American Privacy Emergency
- Authors
- Name
- DeepHorus
The U.S. Department of Commerce issued Directive DAO 216-26 on June 4, 2026, banning modern confidentiality protection techniques in data publications by the Census Bureau and Bureau of Economic Analysis. This order, described by experts as an 'American privacy emergency,' restricts disclosure avoidance to outdated methods from the 1970s, threatening both data utility and subject privacy. The directive bypasses standard administrative procedures and fulfills promises from the Heritage Foundation's Project 2025. Critics argue it is driven by political interests rather than scientific merit, specifically aiming to undermine protections that prevent reidentification of individuals in census data. According to Cynthia Dwork, a Harvard professor and differential privacy pioneer, the order turns its back on over half a century of progress in privacy protection.
DAO 216-26 bans differential privacy, noise infusion, and swapping—techniques used for decades to enable granular data sharing. It only permits coarsening, described as reducing detail through rounding, aggregating, or ranges, and suppression as a last resort. The order forbids 'noise infusion,' which includes adding random values to datasets, a method invented to meet increasing demand for granular data while complying with confidentiality laws. Coarsening and suppression were satisfactory for national aggregates but failed for fine geographic or industrial detail. The ban affects major statistical series like the Quarterly Workforce Indicators, which has used input noise infusion since 2002, and swapping for decennial census publications since 1990.
Without modern privacy techniques, the Census Bureau may have to publish less detailed or fewer statistics, weakening protections. The order could reduce public trust and response rates, as confidentiality is legally required under the Census Act (13 U.S. Code Section 9) and critical for ensuring participation. The Center for Renewing America, an organization founded by OMB Director Russell Vought, has been upfront about the stakes, noting that differential privacy blocks ascertaining citizenship status of individuals. This political motivation undermines the scientific integrity of federal statistics.
Cynthia Dwork, along with over a dozen leading computer scientists and statisticians, cosigned a guest post condemning the order. They call on the scientific community to speak out, emphasizing that DAO 216-26 bypasses legally required administrative procedures and ignores half a century of progress. The post mobilizes researchers to oppose what they see as a politically motivated attack on data privacy and statistical quality.
Legal challenges may arise due to bypassed procedures, but the directive is currently in effect. Researchers urge Congress and the administration to reverse the order to preserve data quality and privacy. The future of statistical data sharing in the U.S. hangs in the balance, as the order could set a precedent for other agencies. Without modern disclosure avoidance, the nation risks losing the ability to produce essential economic and demographic statistics at fine granularity while protecting individual privacy.
