In December 2017 the New York City Council passed Int. 1696, later Local Law 49 of 2018. An earlier draft of the bill would have required city agencies to publish the source code of algorithms used in individual decisions; that requirement was dropped in favour of convening a mayoral task force to study the issue instead. Mayor Bill de Blasio convened the Automated Decision Systems (ADS) Task Force in May 2018 — the first such government body in the United States.
Over 18 months, the Task Force, made up of city officials and outside experts, tried and failed to agree on what should count as an "automated decision system." The statutory definition was broad enough to arguably include a spreadsheet with a summation formula; some members wanted to keep it broad to avoid missing risky tools, others wanted to narrow it to make oversight workable, and no consensus alternative was reached. Its final report, published in November 2019, sets out general principles and process recommendations but names no specific city systems and contains no public inventory of the automated tools New York City agencies actually use.
A December 2019 "shadow report," Confronting Black Boxes, published by the AI Now Institute with the ACLU, the New York Civil Liberties Union, the Brennan Center for Justice and other groups, documented that city agencies resisted even basic disclosure during the process. One participant, Albert Fox Cahn of the Surveillance Technology Oversight Project, described "delays, and obfuscations, and then, by the spring of 2019, outright denials." The shadow report's first recommendation to the New York Police Department was that it "must take a public inventory of the ADS systems it uses" — evidence that, more than eighteen months after the process began, the city's own police department still had not done so.
Included here as a cautionary case: a legally mandated transparency process, without an enforceable requirement to disclose specific systems, produced reports and general recommendations while leaving the underlying opacity of government AI use largely intact.
Read the full analysis: https://www.nyc.gov/site/adstaskforce/index.page
Where this practice's information was retrieved from, and when.