Here is what the poll numbers actually mean in plain terms. When 95% of workers say a human should make the final call on employment decisions, they are not making a philosophical statement. They are reacting to something that is already happening to people they know.
Employers across industries are using AI tools to screen out job applicants before any human reads a resume, to generate productivity scores from keystrokes and mouse movements, and in some cases to trigger disciplinary actions or terminations. In 2024, AI hiring tools processed over 30 million applications in the US while generating hundreds of discrimination complaints. A class action lawsuit against HR software company Workday alleged its AI screening tool systematically disadvantaged Black, older, and disabled applicants. Courts allowed the case to proceed.
This is the context behind the poll numbers. Workers are not afraid of a hypothetical future. They are describing their present.
The AFL-CIO has moved from general concern to a formal agenda. In October 2025, it launched what it calls the Workers First Initiative on AI, a blueprint that calls for human review of all automated employment decisions, the right for workers to appeal AI-driven outcomes, and a ban on using AI purely as a surveillance tool to monitor or penalize staff. The federation represents 63 unions and roughly 15 million workers in the US.
Unions have already started converting these principles into actual contract language. Hollywood writers and actors won protections in 2023 limiting how studios can use AI to replicate their work and likenesses. Casino workers in Las Vegas secured severance of $2,000 per year of service if AI eliminates their role. Microsoft agreed to give workers at its ZeniMax game studio advance notice any time AI adoption could affect their jobs. These are not symbolic wins. They are enforceable terms.
The problem is that most workers are not covered by a union contract. In the US, only about 11% of workers have union representation. That leaves the vast majority of the workforce with no formal mechanism to even find out how AI is being used to evaluate them, let alone challenge it.
Legislation is filling some of that gap, but unevenly. Colorado, Illinois, and New York City all now have laws requiring some form of human oversight or bias auditing when AI tools are used in hiring or employment decisions. California passed rules in 2025 requiring that any automated employment decision system must have a trained person who can override it. The EU goes further: employment-related AI is classified as high-risk under its AI Act, with fines up to 35 million euros or 7% of global turnover for violations.
At the federal level in the US, almost nothing has moved. The Trump administration's AI policy has taken the opposite direction, pushing for fewer constraints on AI development and resisting state-level protections. The AFL-CIO and 32 other labor organizations fought to remove a provision from a federal bill that would have blocked states from enforcing their own AI worker protections for five years. That provision was eventually stripped out.
The business reality underneath all of this is worth noting plainly. Forrester research found that 55% of employers who laid off workers to deploy AI later regretted it. Klarna replaced 700 employees with AI, saw customer service quality decline, and had to bring humans back. Amazon's checkout-free store technology, marketed as AI-powered, was found to rely on hundreds of human workers in India watching camera footage. Companies are making large, fast bets on AI replacing human judgment, and a meaningful number of those bets are not paying off.
For any organization that manages people, the direction of travel is now clear. Workers want a say before AI tools are introduced, transparency about how AI is being used to evaluate them, and a human being accountable for decisions that affect their employment. Governments in Europe and several US states are now codifying exactly that as law. The question is not whether these protections will become standard. It is whether employers get ahead of that shift or get dragged into it.