This bill would add a new article to the state technology law to regulate the use of automated decision-making tools by New York government agencies. It defines covered tools broadly to include software using algorithms, computational models, or artificial intelligence to automate, support, or replace human decision-making, while excluding basic office software and certain internal administrative tools that do not materially affect individual rights, benefits, safety, or welfare. The bill also excludes automated employment decision-making tools already covered elsewhere in law.
The bill requires state agencies that use such tools to publicly disclose them on their websites, including descriptions, when use began, and the purpose of the tool. It also requires agencies seeking to use these tools with meaningful human review to complete recurring impact assessments, at least every two years and before material changes, addressing objectives, design and training data, accuracy, fairness, bias, discrimination, cybersecurity, privacy, public health and safety risks, misuse, and notice to affected individuals. If an assessment finds discriminatory or biased outcomes, the agency must stop using the tool and any information produced by it.
The bill further requires impact assessments to be submitted to the governor and legislative leaders before implementation and posted publicly, subject to redactions for public safety, privacy, or information-security concerns. It also directs the Office of Information Technology Services to maintain a statewide inventory of state automated decision-making tools on the open data website, issue guidance to agencies, and provide the inventory to the governor and legislature. Agencies must also provide a separate disclosure of existing tools to the legislature within one year after the effective date.
The bill’s impact on state law would be significant for state and local agencies, school districts, public authorities, and other public entities that use AI or algorithmic systems in decision-making. It creates new transparency, reporting, and oversight obligations, and it could require agencies to halt use of tools found to produce biased or discriminatory results. It also creates a new statewide inventory and disclosure framework that would make government use of automated systems more visible to policymakers and the public.
The general sentiment reflected in the available voting history is favorable, with the Assembly Science and Technology Committee advancing the bill 8-1 and the Assembly Rules Committee approving it unanimously 28-0. There are no committee transcripts provided, so the discussion record does not show detailed debate. The main points of potential contention are likely the breadth of the disclosure requirements, the burden of recurring impact assessments, and the possibility that agencies may need to discontinue tools if bias or discrimination is found, balanced against privacy, cybersecurity, and public-safety redaction protections.
The bill amends the state technology law by creating a new article on automated decision-making in government agencies and adding a new section requiring a statewide inventory of state automated decision-making tools. It imposes disclosure, impact assessment, reporting, and public posting requirements on state agencies and other government entities using covered tools, and it authorizes redactions for security, privacy, and public-safety reasons. It also requires agencies to disclose existing tools to the legislature and directs the Office of Information Technology Services to collect and publish inventory information.
Available voting history suggests the bill was received positively in committee. It passed the Assembly Science and Technology Committee by an 8-1 vote and the Assembly Rules Committee unanimously by 28-0. No transcript excerpts were provided, so there is no detailed record of floor or committee debate, but the strong committee support indicates broad agreement with the bill’s transparency and oversight goals.
The likely areas of contention are the administrative burden on agencies, the scope of what counts as an automated decision-making tool, and the operational consequences of mandatory impact assessments and public disclosures. Critics may be concerned that the bill could slow deployment of useful technology or expose sensitive information, while supporters are likely focused on accountability, bias prevention, and public transparency. The bill addresses some of those concerns by allowing redactions for security, privacy, and public-safety reasons, but the requirement to stop using tools found to produce discriminatory or biased outcomes remains a significant enforcement feature.