A New York statute originally framed as an AI accountability measure has evolved into a workforce protection law that prevents nearly every public employer in the state from reassigning tasks currently performed by civil servants to automated systems, according to a new Manhattan Institute analysis published in September 2026. The Legislative Oversight of Automated Decision-Making in Government Act, known as the LOADinG Act, now prohibits state agencies, school districts, public universities, counties, cities, towns, and villages from using AI in ways that displace employees, transfer their existing duties, or reduce their compensation—even when no one loses a job or wages.
The law's transformation unfolded across four separate enactments between December 2024 and February 2026. Workforce protections were added just three days before the bill's initial passage in June 2024, expanding from state agencies alone to eventually cover 452,298 municipal and school district employees statewide. Meanwhile, the statute's original accountability and transparency provisions—including impact assessments and meaningful human review requirements—were enacted publicly with delayed effective dates, then repealed through closed-door negotiations before ever taking effect. The report notes that the LOADinG model is already spreading beyond government, with proposed legislation extending similar protections to private-sector workers and imposing a 2% income surcharge on corporations that displace employees through AI.
The report argues that the statute's prohibition on transferring duties "currently performed" by employees applies whether or not anyone loses employment, restricting the reassignment of work itself rather than just its employment consequences. Public-sector unions celebrated the law explicitly as job protection: the Professional Staff Congress stated it "worked closely with lawmakers to draft the language" and said the amended statute "will protect jobs at CUNY as public-sector employers embrace artificial intelligence." The Civil Service Employees Association filed an improper-practice charge in Nassau County in September 2025 alleging that automated systems had taken over case-input work historically handled by union members, pairing the legal filing with a "Rally Against the Bot" the following month.
The restrictions map directly onto AI's value proposition for government, according to the Manhattan Institute report. Emerging research shows generative AI raised customer-support productivity by 14% in a Stanford and MIT study of more than 5,000 agents, cut routine writing time by roughly 40% in a randomized trial, and saved U.K. civil servants a self-reported 26 minutes per day—nearly two weeks per employee annually. Researchers at the Alan Turing Institute estimate that 84% of the U.K. central government's complex but repetitive transactions, about 143 million each year, are highly automatable, and saving just one minute per transaction would free the equivalent of 1,200 person-years of work annually. The report notes that federal agencies disclosed roughly 3,600 AI use cases in 2025, a nearly 70% jump from the previous year. But the statute's broad language creates uncertainty for managers: key terms like "displacement" and "duties currently performed" remain undefined, and no court or agency has interpreted them, so the cost of guessing wrong falls entirely on the public employer while the benefits of deployment are diffuse.
The report concludes that the LOADinG Act inverts government's core purpose by treating the preservation of existing work arrangements as an end that public administration must serve, even at the cost of slower, costlier, and worse services. The law's restrictions traveled from New York's state agencies to its school districts, public universities, and local governments in 20 months, and the model is now positioned to spread to other states and the private economy. The predictable result, according to the analysis, is that agencies will confine AI to peripheral uses that don't touch work assignments—permitting the technology everywhere except where it's most valuable. The question the statute poses is whether American governments will approach artificial intelligence as an instrument for serving citizens better or as a threat to be contained on behalf of those they employ.