Can My Employer Make Me Use AI?

The question came up in a People Also Ask box, which is to say: people are typing it into Google. They want a real answer. They have not been given one.

If you search "AI governance" or "AI policy for employees," the page that comes back is for the person who is going to make you use the AI. The templates are written for the executive who has decided that the company will adopt an AI tool and now needs a policy document to attach to the rollout. The frameworks are written for the board that wants reassurance about compliance exposure. The implementation guides are written for the HR director who is operationalizing the executive's decision. There is no shortage of material. SHRM, Deloitte, the major law firms, the Big Four — all of them have a position paper on AI governance. None of those position papers is written for you.

You, in this construction, are the person being governed. You are the variable in someone else's framework. The question of whether you have any say in the matter — whether you can decline, what happens if you do, what your employer is and isn't allowed to require of you — that question is not what the AI governance industry produces content about. So you go to Google with the actual question, and Google's algorithm surfaces it as a "People also ask" item under the executive-facing material, and then offers you the executive-facing material as the answer.

This piece is an attempt at the answer that Google did not surface.

What the law actually says (in the United States, in 2026)

The short version: in most US states, your employer can require you to use the tools they provide, including AI tools, as a condition of continued employment. The doctrine that makes this possible is called employment at will, and it is the default in 49 of 50 states. (Montana is the exception, and even there the protection is partial.) Under at-will employment, your employer can change the terms and tools of your job, and if you refuse the changes, the employer can terminate the relationship. They don't owe you a reason. You don't owe them notice.

This is the legal floor. Above it sit a number of structures that complicate the picture.

If you work under a union contract, the introduction of new technology — including AI tools that materially change your job — may be a mandatory subject of bargaining under the National Labor Relations Act. Your employer cannot unilaterally impose AI tools on you if the contract reserves that decision to the bargaining process. The Communications Workers of America, the Writers Guild, the actors' union SAG-AFTRA, and several public-sector unions have negotiated specific AI clauses into their contracts in the last two years. If you have a union, this is the place to look first.

If you have a written employment contract (most US workers do not; some do), the contract's terms govern. Some contracts restrict the employer's ability to change job duties. Most do not.

If you are a public-sector employee, you may have additional procedural rights that govern how technology changes are implemented — civil service protections, due process for adverse employment actions, sometimes a meet-and-confer requirement before changes take effect.

If the AI tool would require you to do something that violates your professional ethical obligations — a lawyer being asked to use an AI tool that compromises client confidentiality, a clinician being asked to use a tool that the FDA has not cleared for the clinical use case, a journalist being asked to publish AI-generated content as their own work — there are sector-specific obligations that may override the employer's directive. Whistleblower protections exist for narrow categories of conduct; AI deployment that violates safety regulations or commits fraud may fall under them.

If the AI tool, in its operation, would discriminate against you on the basis of a protected characteristic — race, sex, age over 40, disability, national origin, religion, pregnancy — Title VII of the Civil Rights Act and the ADA still apply. The fact that the discrimination is operationalized by an algorithm does not extinguish the cause of action. The EEOC issued specific guidance on this in 2023 and has updated it twice since. If you believe an AI tool your employer requires you to use is discriminating against you, the EEOC charge process is the formal route.

If you live in the European Union, the answer is meaningfully different. The EU AI Act, in force since 2024, classifies certain employer-facing AI uses (hiring evaluation, performance assessment, surveillance) as high-risk, with transparency, human-oversight, and worker-consultation requirements. Several EU member states have additional national-level worker protections. If you are in the EU, the question "can my employer make me use AI" has a more textured answer than it does in the US.

If you live in California, Colorado, or Illinois, state-level legislation specific to AI in employment is either in force or imminent. California's AB 2930 (introduced) and SB 1047 (in flux) address aspects of this. Illinois has had its Artificial Intelligence Video Interview Act since 2020. Colorado's AI Act, passed in 2024, includes employment provisions taking effect in 2026.

That is the legal landscape, compressed. None of it is what you wanted to hear. Most of it confirms that the answer to the headline question, in most US workplaces, is yes — your employer can require AI use, and if you refuse, the employer's recourse is generally not constrained.

What the question is actually asking

The legal answer is not the whole answer. People typing this question into Google are not, in most cases, looking for the at-will employment doctrine. They are looking for something the law does not have a clean category for.

They are asking: do I have any standing to refuse?

They are asking: if the AI tool changes the meaning of my work in ways I object to, does my objection count?

They are asking: if my employer's adoption of AI is bad for me — bad for my expertise, bad for my professional development, bad for my ability to do the work the way I think it should be done, bad for the people I serve — is there anywhere to go with that?

The legal answer to all three of these is mostly no. The structural answer is more complicated.

The compliance template and the worker

Here is what the AI governance industry actually does. A consulting firm writes a template for "responsible AI use." The template includes sections on data privacy, on bias mitigation, on human oversight, on incident response. The template is rigorous in form. An organization adopts the template. The template is now policy. The policy specifies that AI tools will be used in accordance with ethical principles, that human oversight will be maintained, that incidents will be reported and reviewed.

What the template does not specify is the operational question that determines whether the policy means anything: when a worker raises a concern about an AI tool — its accuracy, its appropriateness for the use case, its effect on the worker's ability to do good work — what is the procedure for that concern to receive a substantive response, and who is accountable if it does not?

The templates do not contain this procedure. They cannot. The procedure would constrain the executive's discretion to deploy the tool, and the template is written for the executive. Including the procedure would make the template harder to sell.

This is the structural gap. The AI governance industry produces compliance documents that are formally rigorous and operationally hollow. The worker reading the policy that governs their use of AI tools is reading a document that does not contain the answer to the question they actually want answered.

When you ask "can my employer make me use AI," you are asking, in part, whether the policy your employer just adopted means anything for you. The honest answer is that it probably doesn't. The policy was written for compliance posture, not for your standing.

What might actually work

This part is harder, and the data does not support strong recommendations. Some general observations:

Document everything. If an AI tool produces an output you believe is wrong, save the output, save the prompt, save the date, save your assessment of why it is wrong. Do this even if no one asks for it. The documentation creates a record that can be referenced later, in a grievance, in a regulatory complaint, in a lawsuit, in a journalist's investigation, in your own decision-making about whether to stay in the role.

If you are a clinician, a lawyer, an accountant, a licensed engineer, or any other professional with a regulatory body — the regulatory body is where to go when the employer requires use of a tool that conflicts with your professional obligations. Your license is the leverage. Most regulatory bodies have not yet issued specific AI guidance; some have. Check.

If you are in a workplace with a union or with a credible path to organizing one, the AI question is one of the most useful organizing topics available. Workers across very different jobs share an interest in not having decisions about their work imposed by tools they did not choose and cannot opt out of. The Communications Workers, the National Nurses United, the Teamsters, and several locals have active AI-deployment campaigns. Find the one closest to your work.

If you are in a non-union workplace and the AI tool's deployment is causing measurable harm — to patients, to clients, to students, to the public — the harm is reportable. State attorneys general, sector regulators, and journalists are all currently interested in AI-deployment failures. The reporting is not without risk to your job; the risk is real and you should know about it before you make the call. The harm is also real, and the reporting is sometimes how the harm gets addressed.

If none of these apply — if you are at-will, non-union, in a sector without a strong regulator, and the harm from the AI tool is more diffuse than reportable — you are in the largest category of US workers, and the honest answer is that your standing is limited. You can refuse and accept the consequences. You can comply and document your objections. You can leave. You can talk to your coworkers about whether they have the same objections, which is the precondition for any collective response. Those are the levers.

What this register is missing

The AI governance literature, the way it is written today, gives executives the language to govern AI deployment. It does not give workers the language to engage AI deployment as something happening to them, by them, for purposes they may or may not share. The literature gives the people deploying the AI a vocabulary that the people being deployed-against do not have.

This is not an accident. The AI governance literature is produced by consulting firms, law firms, and standards bodies whose clients are the deployers, not the deployed-against. The vocabulary follows the money.

The question "can my employer make me use AI?" is not a niche question. It surfaces in People Also Ask boxes because millions of workers are typing it into Google. The supply of substantive answers is shaped by who pays for content production. Adjustments to that supply will not come from the consulting firms. They will come from somewhere else.

This publication is one place. It is not the only one, and it is not enough. But the gap is real and the question deserves an answer that is not written for the person who is going to deploy the tool.

The Foragentis team publishes The Human and I to engage the questions about AI that the available answers do not address. If your organization is making decisions about AI deployment and wants to engage the worker-perspective seriously — including substantive consultation processes that go beyond compliance posture — Foragentis' organizational consulting practice works on exactly this. The 90-day discovery engagement starts at the question your workers are actually asking.