# Mandatory AI use and accommodation risk[^about] How mandatory AI use affects essential job functions, ADA and Title VII accommodations, review processes, and workplace AI tool design. ## Is mandatory AI use itself an essential job function? {#is-mandatory-ai-use-an-essential-job-function} **Short answer.** Sometimes, but only when the record shows the job is genuinely built around using the AI workflow. The safer analysis separates the essential work output from the employer's preferred tool or interface. The law is mostly old. A mandatory AI-use rule is usually not a new legal category; it is a work rule, interface choice, performance metric, or testing method that sits inside existing accommodation law. Under the ADA, the hard question is whether use of a specific AI tool is actually an essential function or only the employer's preferred method of performing an essential function, and whether another effective path exists without "significant difficulty or expense"[^42-u-s-c-section-12111-10]. Under Title VII after *Groff v. DeJoy*, the question is whether a sincere religious objection can be accommodated without substantial increased costs in relation to the conduct of [the employer's] particular business. State regimes like FEHA, NYSHRL, NYCHRL, and New Jersey's 2025 guidance make the same point in a more operational way: once AI use becomes mandatory, rigid enforcement becomes harder to defend if the tool itself blocks accommodation or if the same output can be reached through another workflow. [^42-u-s-c-section-12112-b-5-a][^42-u-s-c-section-12111-10][^42-u-s-c-section-2000e-j][^groff-v-dejoy-600-u-s-447-2023][^california-government-code-section-12940-n][^new-york-executive-law-section-296][^new-york-city-administrative-code-section-8-107][^new-jersey-division-on-civil-rights-guidance-on] The ADA fits the problem without much doctrinal stretching. Discrimination includes "not making reasonable accommodations to the known physical or mental limitations"[^42-u-s-c-section-12112-b-5-a] of an otherwise qualified applicant or employee, unless the employer can show undue hardship. The statute also says reasonable accommodation may include modified schedules, reassignment, equipment or device modification, and modifications of examinations, training materials, or policies. The implementing regulation adds that it may be necessary to begin an "informal, interactive process"[^29-c-f-r-section-1630-2-o-3] to identify the limitation and a workable accommodation. [^42-u-s-c-section-12112-b-5-a][^42-u-s-c-section-12111-10][^29-c-f-r-section-1630-2-o-3] That matters because the employer can insist on the essential function of the job, but it does not automatically get to freeze the current vendor, interface, or workflow if another effective means exists. Current federal AI materials already assume the accommodation framework applies. The EEOC's 2024 handout for workers says existing employment laws may require reasonable accommodation even when the employer uses AI, and it describes AI use across recruiting, video interviews, pay, promotion, productivity monitoring, layoffs, and termination. The EEOC's 2022 AI-and-ADA technical assistance and its 2023 visual-disabilities guidance point the same way, and DOJ's 2022 hiring guidance treats automated tools as fully inside disability-discrimination law rather than outside it. [^eeoc-employment-discrimination-and-ai-for-worker][^eeoc-artificial-intelligence-and-the-ada][^eeoc-visual-disabilities-in-the-workplace-and-th][^doj-algorithms-artificial-intelligence-and-disab] The immediate consequence is that a company cannot rely on the novelty of the tool to argue that accommodation rules have not caught up. The state analogues make the procedural layer more explicit. FEHA separately prohibits failure to provide reasonable accommodation and failure to engage in a "timely, good faith, interactive process"[^california-government-code-section-12940-n]. Its religious-accommodation section asks whether the employer explored any available reasonable alternative means of accommodating and uses a `significant difficulty or expense` hardship concept. [^california-government-code-section-12940-n][^california-government-code-section-12926] New York State bars employers from imposing terms that require a person to "violate or forego"[^new-york-executive-law-section-296] a sincerely held religious practice. New York City goes further, requiring a "cooperative dialogue"[^new-york-city-administrative-code-section-8-107] and a written final determination for disability and religion requests. [^new-york-executive-law-section-296][^new-york-city-administrative-code-section-8-107][^nyc-commission-on-human-rights-title-8-civil-rig] New Jersey's January 2025 guidance is the most AI-specific state source in the set. The Division on Civil Rights says algorithmic discrimination can violate the LAD when it "precludes or impedes"[^new-jersey-division-on-civil-rights-guidance-on] reasonable accommodations for disability or religion. Its examples are concrete: a typing-speed tool that cannot read input from a nontraditional keyboard, or a productivity tool that flags accommodated breaks because the system was never configured for them. [^new-jersey-division-on-civil-rights-guidance-on] That is perhaps the clearest existing public statement that AI accommodation risk may appear in the measurement layer, not just in the chatbot interface. The employment bar is unusually aligned on the disability side. Littler's 2022 ADA memo reads the EEOC position straightforwardly: when disability makes an AI tool hard to use or drags down the result, the accommodation duty stays with the employer, even if the practical problem began with a vendor product. Jackson Lewis took the same event and emphasized accessible design, accessible instructions, and clear accommodation channels rather than any special AI exception. [^littler-eeoc-issues-guidance-on-artificial-intel][^jackson-lewis-eeoc-doj-release-expectations-on-e] The non-obvious point is how little space the firms give to the idea that vendor failure somehow interrupts the employer's own obligations. More recent commentary extends that logic from hiring tools to ordinary workplace AI mandates. Ogletree's June 2025 overview treats reasonable accommodation as a core employment-law issue for workplace AI, not a narrow recruiting issue. Its April 2026 piece on religious objections is even more revealing because it treats the issue as current workplace fact: some employers are already receiving accommodation requests from workers who say AI use conflicts with religious belief. [^ogletree-deakins-the-intersection-of-artificial][^ogletree-deakins-the-mark-of-the-bot-when-employ] That does not mean every objection qualifies as religious. It means the employer bar no longer sees the question as hypothetical. The California and New Jersey commentary pushes in the same direction. Jackson Lewis and Littler both read California's 2025 AI employment regulations to mean that employers may have to accommodate AI-enabled barriers where the system measures traits like dexterity, reaction time, tone of voice, or facial expression. Littler and Ogletree read New Jersey's 2025 guidance similarly: an automated system can create accommodation exposure when it blocks disability or religious accommodation rather than merely screening for bias in the abstract. [^jackson-lewis-california-s-new-ai-regulations-ta][^littler-california-approves-landmark-ai-employme][^littler-ai-in-the-garden-state-new-guidance-on-a][^ogletree-deakins-new-jersey-division-on-civil-ri] Seyfarth's treatment of the Department of Labor's 2024 promising practices lands in the same place from a different angle. AI systems are expected to be accessible and to leave room for accommodation requests. Across the firms, there is very little daylight on the main issue. The disagreement is mostly at the edges: which requests are sincere, which burdens are substantial, and whether a given AI workflow is genuinely central to the job or just the current management preference. [^seyfarth-shaw-department-of-labor-issues-compreh][^ogletree-deakins-the-mark-of-the-bot-when-employ] The important shift is from optional tool to employment term. Public memos from Shopify and Box matter less because they are law than because they show how the facts are changing: AI use described as a `fundamental expectation`, a staffing gate, and a redesign of daily work. Once the record looks like that, an accommodation dispute stops being about experimentation with a new tool and starts looking like any other dispute over a required method, review metric, or condition of employment. [^the-verge-shopify-ceo-says-no-new-hires-without][^box-ai-first-transformation-box-s-principles-str] This is where the essential-function question becomes real. A company may have a stronger story when the role is itself about supervising model outputs, operating inside a shared AI workflow, or maintaining a specific AI-enabled process. The story is weaker when the essential function is still the underlying legal analysis, coding, drafting, selling, or project management, and AI is mainly the preferred path to that output. The ADA lets employers hold output and production standards constant. It does not automatically treat every preferred method as untouchable. [^42-u-s-c-section-12112-b-5-a][^42-u-s-c-section-12111-10][^eeoc-enforcement-guidance-on-reasonable-accommod] Perhaps yes where the role is genuinely built around operating inside a named AI workflow, and the written description and practice record say so. Perhaps not where the real essential function is still the underlying work product and AI is only the preferred route to it. [^42-u-s-c-section-12111-10][^42-u-s-c-section-12112-b-5-a][^eeoc-enforcement-guidance-on-reasonable-accommod] ## How far can religious objections to mandatory AI use go? {#how-far-can-religious-objections-to-ai-use-go} **Short answer.** They can go farther than minor inconvenience, but only for sincere religious conflicts rather than secular objections to new technology. After *Groff*, the employer needs a concrete undue-hardship analysis tied to its actual business. Title VII is less AI-specific but no less relevant. The statute says religion includes "all aspects of religious observance and practice, as well as belief"[^42-u-s-c-section-2000e-j-2] unless the employer cannot reasonably accommodate without undue hardship. After *Groff*, the old `more than de minimis` shorthand no longer does the work; the burden has to rise to substantial increased costs in the context of the particular business. The pre-*Groff* regulation still helps on the shape of possible alternatives, including voluntary substitutes, schedule changes, and transfer, but not on hardship threshold. [^42-u-s-c-section-2000e-j-2][^groff-v-dejoy-600-u-s-447-2023-2][^29-c-f-r-section-1605-2][^eeoc-religious-discrimination] So an AI mandate challenged on religious grounds is not decided by mild workflow friction or a generalized preference for uniformity. Religious-objection disputes may be where first impression shows up fastest. Title VII does not protect a generalized ethical dislike of AI, a secular professionalism argument, or a preference not to use new software. But after *Groff*, it also does not let an employer stop at minor friction. The real fight is likely to be over sincerity, whether the objection is actually religious, and whether the requested exception changes the business in a substantial way or merely complicates a preferred workflow. [^42-u-s-c-section-2000e-j-2][^groff-v-dejoy-600-u-s-447-2023-2][^eeoc-questions-and-answers-religious-discriminat][^ogletree-deakins-the-mark-of-the-bot-when-employ-2] We think farther than the old `de minimis` shorthand suggested, because the burden now has to be substantial in the context of the actual business. But Title VII still does not protect objections that are political, economic, or merely personal rather than religious. [^42-u-s-c-section-2000e-j-2][^groff-v-dejoy-600-u-s-447-2023-2][^eeoc-religious-discrimination][^eeoc-questions-and-answers-religious-discriminat] ## What accommodations work when employees cannot use mandatory AI tools? {#what-accommodations-work-for-mandatory-ai-tools} **Short answer.** Common options include accessible settings, alternate tools, different input methods, more time, training, human review, or removing the employee from an AI-use metric. The employer can still preserve essential functions and production standards. The harder cases may come from instrumentation rather than from the chatbot prompt box. A worker whose accommodation already includes extra break time, slower input, assistive devices, or alternative input modes can look deficient to an AI productivity system that was never configured for those facts. New Jersey's guidance makes this explicit. So companies that turn AI usage into a KPI, utilization score, or performance-review factor inherit accommodation risk at the measurement layer as well as at the tool layer. [^new-jersey-division-on-civil-rights-guidance-on-2][^eeoc-employment-discrimination-and-ai-for-worker-2] Perhaps an accessible version of the same tool, a different tool, different input or output modes, more time or training, a human-reviewed alternative, or removal from an AI-usage metric while keeping the substantive output standard. The counterargument is familiar too: accommodation does not erase essential functions or reduce production standards, and reassignment remains a last resort under the ADA. [^42-u-s-c-section-12111-10-2][^42-u-s-c-section-12112-b-5-a-2][^29-c-f-r-section-1630-2-o-3-2][^eeoc-artificial-intelligence-and-the-ada-2] ## How should employers review suspicious mandatory AI accommodation requests? {#how-should-employers-review-suspicious-ai-accommodation-requests} **Short answer.** They can ask targeted questions about disability-related need or religious sincerity, but they should still run the ordinary accommodation process. Treating a doubtful request as insubordination too quickly creates avoidable process risk. The documented record changes the downside more than slogans do. When the file shows that AI use was identified as essential before the dispute, the employer tested accessible configurations or alternative workflows, requested only limited supporting material, and analyzed burden in concrete business terms, the dispute looks like an accommodation process honestly run. When the file shows only a broad AI-first norm, a vendor chosen without attention to accessibility, and a refusal to consider another effective workflow, the same policy looks much more brittle. The ADA separately narrows certain damages where the employer can show "good faith efforts, in consultation with the person with the disability"[^42-u-s-c-section-1981a-a-3] to identify and make an equally effective accommodation. [^42-u-s-c-section-1981a-a-3][^eeoc-enforcement-guidance-on-reasonable-accommod-2][^29-c-f-r-section-1630-2-o-3-3] The ADA permits limited documentation about disability and need, and Title VII permits inquiry into sincerity. But the source set points toward the same conclusion on both sides: even a doubtful request still tends to trigger the ordinary accommodation process rather than immediate treatment as insubordination. [^eeoc-enforcement-guidance-on-reasonable-accommod-2][^eeoc-questions-and-answers-religious-discriminat-2][^ogletree-deakins-the-mark-of-the-bot-when-employ-3] ## Can an AI tool itself be a workplace accommodation? {#can-an-ai-tool-be-a-workplace-accommodation} **Short answer.** Yes, if the tool's accessibility features help the employee perform the essential work. The same tool can also create risk if it breaks assistive technology, misreads accommodated performance, or leaves no accessible alternative. Mandatory AI use is not automatically anti-accommodation. Some AI products can expand access. Microsoft publishes screen-reader guidance for Copilot in Word, Outlook, and PowerPoint. Google publishes Workspace accessibility materials describing screen-reader support, keyboard shortcuts, braille-display compatibility, and voice typing. That cuts in both directions. It weakens the idea that every disability-based objection deserves a full opt-out, but it also makes vendor choice harder to treat as neutral. If comparable products have accessible pathways and the chosen workflow does not, the accessibility gap becomes part of the employer's facts. [^microsoft-support-use-copilot-in-word-with-a-scr][^microsoft-support-use-copilot-in-outlook-with-a][^microsoft-support-basic-tasks-using-a-screen-rea][^google-accessibility-google-workspace-for-divers][^google-accessibility-help-google-workspace-admin] Perhaps yes when the tool's voice, reading, summarization, or keyboard features are the reason the employee can perform essential functions at all. The opposite possibility is that the same tool becomes the problem once it breaks assistive technology, misreads accommodated performance, or is locked into a mandatory workflow that has no accessible alternative. [^microsoft-support-use-copilot-in-word-with-a-scr][^google-accessibility-help-google-workspace-admin][^new-jersey-division-on-civil-rights-guidance-on-3] [^about]: By Steven Obiajulu, J.D. Published by [openagreements.org](https://openagreements.org). Last reviewed 2026-04-20. License: CC BY 4.0. Steven Obiajulu, J.D. edits this topic article for Federal + multi-state coverage. It synthesizes legal sources and is not legal advice. This article is for informational purposes only and does not create an attorney-client relationship. [^42-u-s-c-section-12111-10]: **42 U.S.C. section 12111(10)** — "significant difficulty or expense" *42 U.S.C. section 12111(10).* [^42-u-s-c-section-12112-b-5-a]: **42 U.S.C. section 12112(b)(5)(A)** — "not making reasonable accommodations to the known physical or mental limitations" *42 U.S.C. section 12112(b)(5)(A).* [^42-u-s-c-section-2000e-j]: **42 U.S.C. section 2000e(j)** — "all aspects of religious observance and practice, as well as belief" *42 U.S.C. section 2000e(j).* [^groff-v-dejoy-600-u-s-447-2023]: **Groff v. DeJoy, 600 U.S. 447 (2023)** — "We hold that showing ‘more than a de minimis cost,’ as that phrase is used in common parlance, does not suffice to establish ‘undue hardship’ under Title VII." *Groff v. DeJoy, 600 U.S. 447 (2023).* [^california-government-code-section-12940-n]: **California Government Code section 12940(n)** — "timely, good faith, interactive process" *California Government Code section 12940(n).* [^new-york-executive-law-section-296]: **New York Executive Law section 296** — "violate or forego" *New York Executive Law section 296.* [^new-york-city-administrative-code-section-8-107]: **New York City Administrative Code section 8-107** — "cooperative dialogue" *New York City Administrative Code section 8-107.* [^new-jersey-division-on-civil-rights-guidance-on]: **New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination** — "precludes or impedes" *New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination.* [^29-c-f-r-section-1630-2-o-3]: **29 C.F.R. section 1630.2(o)(3)** — "informal, interactive process" *29 C.F.R. section 1630.2(o)(3).* [^eeoc-employment-discrimination-and-ai-for-worker]: **EEOC, Employment Discrimination and AI for Workers** — "Federal employment discrimination laws protect you when AI systems are used to discriminate against you on the basis of your race, color, religion, sex (including gender, sexual orientation, and pregnancy), national origin, age (40 or older), disability, or genetic information." *EEOC, Employment Discrimination and AI for Workers.* [^eeoc-artificial-intelligence-and-the-ada]: **EEOC, Artificial Intelligence and the ADA** — "The Americans with Disabilities Act and the Use of Software, Algorithms, and Artificial Intelligence to Assess Job Applicants and Employees" *EEOC, Artificial Intelligence and the ADA.* [^eeoc-visual-disabilities-in-the-workplace-and-th]: **EEOC, Visual Disabilities in the Workplace and the Americans with Disabilitie...** — "An employer may exclude an individual with a vision impairment from a job for safety reasons only when the individual poses a direct threat. A ‘direct threat’ is a significant risk of substantial harm to the health or safety of the individual or others that cannot be eliminated or reduced through reasonable accommodation." *EEOC, Visual Disabilities in the Workplace and the Americans with Disabilities Act.* [^doj-algorithms-artificial-intelligence-and-disab]: **DOJ, Algorithms, Artificial Intelligence, and Disability Discrimination in Hi...** — "An employer who chooses to use a hiring technology must ensure that its use does not cause unlawful discrimination on the basis of disability." *DOJ, Algorithms, Artificial Intelligence, and Disability Discrimination in Hiring.* [^california-government-code-section-12926]: **California Government Code section 12926** — "‘Essential functions’ means the fundamental job duties of the employment position the individual with a disability holds or desires." *California Government Code section 12926.* [^nyc-commission-on-human-rights-title-8-civil-rig]: **NYC Commission on Human Rights, Title 8 Civil Rights** — "It shall be an unlawful discriminatory practice for any person who is the owner, franchisor, franchisee, lessor, lessee, proprietor, manager, superintendent, agent or employee of any place or provider of public accommodation: 1. Because of any person's actual or perceived race, creed, color, national origin, age, gender, disability, marital status, partnership status, sexual orientation, uniformed service, height, weight, or immigration or citizenship status, directly or indirectly: (a) To refuse, withhold from or deny to such person the full and equal enjoyment, on equal terms and conditions, of any of the accommodations, advantages, services, facilities or privileges of the place or provider of public accommodation" *NYC Commission on Human Rights, Title 8 Civil Rights.* [^littler-eeoc-issues-guidance-on-artificial-intel]: **Littler commentary** — "Noting that employers are responsible for ADA-violating outcomes even when a software tool is created or used by a third-party vendor or agent, the Commission provides examples of so-called ‘Promising Practices’ that employers can engage in to demonstrate good-faith efforts to meet ADA requirements." *Littler, EEOC Issues Guidance on Artificial Intelligence and the Americans with Disabilities Act.* [^jackson-lewis-eeoc-doj-release-expectations-on-e]: **Jackson Lewis commentary** — "The EEOC’s TAD applies the Americans with Disabilities Act (ADA), including regulations and existing guidance, where technology intersects with workplace legal issues." *Jackson Lewis, EEOC, DOJ Release Expectations on Employers' Use of Technology (AI) in Employment Decisions.* [^ogletree-deakins-the-intersection-of-artificial]: **Ogletree Deakins commentary** — "Using automated technology to make workforce decisions presents significant legal risks under existing anti-discrimination laws, such as Title VII, the ADEA, and the ADA, because bias in algorithms can lead to allegations of discrimination." *Ogletree Deakins, The Intersection of Artificial Intelligence and Employment Law.* [^ogletree-deakins-the-mark-of-the-bot-when-employ]: **Ogletree Deakins commentary** — "Federal and state laws require employers to provide reasonable accommodations for an employee’s religious beliefs, unless it would impose an undue hardship." *Ogletree Deakins, The Mark of the Bot: When Employees Raise Religious Objections to Workplace AI Usage.* [^jackson-lewis-california-s-new-ai-regulations-ta]: **Jackson Lewis, California's New AI Regulations Take Effect Oct. 1. Here's You...** — "The regulations prohibit employers from using ADS or selection criteria that discriminate against applicants or employees based on protected categories defined under the Fair Employment and Housing Act (FEHA)." *Jackson Lewis, California's New AI Regulations Take Effect Oct. 1. Here's Your Compliance Checklist.* [^littler-california-approves-landmark-ai-employme]: **Littler, California Approves Landmark AI Employment Regulations** — "Revisions to Title 2 of the California Code of Regulations will govern the use of AI-based tools in California starting October 1, 2025." *Littler, California Approves Landmark AI Employment Regulations.* [^littler-ai-in-the-garden-state-new-guidance-on-a]: **Littler commentary** — "The Guidance clarifies that the New Jersey Law Against Discrimination (NJLAD) prohibits ‘algorithmic discrimination,’ that is, discrimination resulting from a covered entity’s use of automated decision-making tools." *Littler, AI in the Garden State: New Guidance on Algorithmic Discrimination and New Jersey Law.* [^ogletree-deakins-new-jersey-division-on-civil-ri]: **Ogletree Deakins commentary** — "the guidance reinforces that employers are fully responsible for the AI technology they utilize and may not delegate their compliance responsibilities to third parties." *Ogletree Deakins, New Jersey Division on Civil Rights Issues New Guidance on Algorithmic Discrimination.* [^seyfarth-shaw-department-of-labor-issues-compreh]: **Seyfarth Shaw commentary** — "OFCCP unambiguously warns that federal contractors cannot delegate or avoid their nondiscrimination and affirmative action obligations by using third-party AI products." *Seyfarth Shaw, Department of Labor Issues Comprehensive Artificial Intelligence Promising Practices Designed to Avoid Bias.* [^the-verge-shopify-ceo-says-no-new-hires-without]: **The Verge, Shopify CEO says no new hires without proof AI can't do the job** — "Before asking for more Headcount and resources, teams must demonstrate why they cannot get what they want done using AI." *The Verge, Shopify CEO says no new hires without proof AI can't do the job.* [^box-ai-first-transformation-box-s-principles-str]: **Box, AI-First Transformation: Box's Principles, Strategy, and Execution Framework** — "Box's AI executive sponsorship is led by Olivia Nottebohm, the company’s COO, Ravi Malick, the CIO, and Jessica Swank, Chief People Officer, to represent business, IT and people aspects — all of which are critical." *Box, AI-First Transformation: Box's Principles, Strategy, and Execution Framework.* [^eeoc-enforcement-guidance-on-reasonable-accommod]: **EEOC, Enforcement Guidance on Reasonable Accommodation and Undue Hardship und...** — "Title I of the Americans with Disabilities Act of 1990 (the ‘ADA’)(1) requires an employer(2) to provide reasonable accommodation to qualified individuals with disabilities who are employees or applicants for employment, unless to do so would cause undue hardship." *EEOC, Enforcement Guidance on Reasonable Accommodation and Undue Hardship under the ADA.* [^42-u-s-c-section-2000e-j-2]: **42 U.S.C. section 2000e(j)** — "all aspects of religious observance and practice, as well as belief" *42 U.S.C. section 2000e(j).* [^groff-v-dejoy-600-u-s-447-2023-2]: **Groff v. DeJoy, 600 U.S. 447 (2023)** — "We hold that showing ‘more than a de minimis cost,’ as that phrase is used in common parlance, does not suffice to establish ‘undue hardship’ under Title VII." *Groff v. DeJoy, 600 U.S. 447 (2023).* [^29-c-f-r-section-1605-2]: **29 C.F.R. section 1605.2** — "Section 701(j) makes it an unlawful employment practice under section 703(a)(1) for an employer to fail to reasonably accommodate the religious practices of an employee or prospective employee, unless the employer demonstrates that accommodation would result in undue hardship on the conduct of its business." *29 C.F.R. section 1605.2.* [^eeoc-religious-discrimination]: **EEOC, Religious Discrimination** — "undue hardship is shown when a burden is substantial in the overall context of an employer’s business" *EEOC, Religious Discrimination.* [^eeoc-questions-and-answers-religious-discriminat]: **EEOC, Questions and Answers: Religious Discrimination in the Workplace** — "Title VII requires an employer, once on notice that a religious accommodation is needed, to reasonably accommodate an employee whose sincerely held religious belief, practice, or observance conflicts with a work requirement, unless doing so would pose an undue hardship." *EEOC, Questions and Answers: Religious Discrimination in the Workplace.* [^ogletree-deakins-the-mark-of-the-bot-when-employ-2]: **Ogletree Deakins commentary** — "Federal and state laws require employers to provide reasonable accommodations for an employee’s religious beliefs, unless it would impose an undue hardship." *Ogletree Deakins, The Mark of the Bot: When Employees Raise Religious Objections to Workplace AI Usage.* [^new-jersey-division-on-civil-rights-guidance-on-2]: **New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination** — "precludes or impedes" *New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination.* [^eeoc-employment-discrimination-and-ai-for-worker-2]: **EEOC, Employment Discrimination and AI for Workers** — "Federal employment discrimination laws protect you when AI systems are used to discriminate against you on the basis of your race, color, religion, sex (including gender, sexual orientation, and pregnancy), national origin, age (40 or older), disability, or genetic information." *EEOC, Employment Discrimination and AI for Workers.* [^42-u-s-c-section-12111-10-2]: **42 U.S.C. section 12111(10)** — "significant difficulty or expense" *42 U.S.C. section 12111(10).* [^42-u-s-c-section-12112-b-5-a-2]: **42 U.S.C. section 12112(b)(5)(A)** — "not making reasonable accommodations to the known physical or mental limitations" *42 U.S.C. section 12112(b)(5)(A).* [^29-c-f-r-section-1630-2-o-3-2]: **29 C.F.R. section 1630.2(o)(3)** — "informal, interactive process" *29 C.F.R. section 1630.2(o)(3).* [^eeoc-artificial-intelligence-and-the-ada-2]: **EEOC, Artificial Intelligence and the ADA** — "The Americans with Disabilities Act and the Use of Software, Algorithms, and Artificial Intelligence to Assess Job Applicants and Employees" *EEOC, Artificial Intelligence and the ADA.* [^42-u-s-c-section-1981a-a-3]: **42 U.S.C. section 1981a(a)(3)** — "good faith efforts, in consultation with the person with the disability" *42 U.S.C. section 1981a(a)(3).* [^eeoc-enforcement-guidance-on-reasonable-accommod-2]: **EEOC, Enforcement Guidance on Reasonable Accommodation and Undue Hardship und...** — "Title I of the Americans with Disabilities Act of 1990 (the ‘ADA’)(1) requires an employer(2) to provide reasonable accommodation to qualified individuals with disabilities who are employees or applicants for employment, unless to do so would cause undue hardship." *EEOC, Enforcement Guidance on Reasonable Accommodation and Undue Hardship under the ADA.* [^29-c-f-r-section-1630-2-o-3-3]: **29 C.F.R. section 1630.2(o)(3)** — "informal, interactive process" *29 C.F.R. section 1630.2(o)(3).* [^eeoc-questions-and-answers-religious-discriminat-2]: **EEOC, Questions and Answers: Religious Discrimination in the Workplace** — "Title VII requires an employer, once on notice that a religious accommodation is needed, to reasonably accommodate an employee whose sincerely held religious belief, practice, or observance conflicts with a work requirement, unless doing so would pose an undue hardship." *EEOC, Questions and Answers: Religious Discrimination in the Workplace.* [^ogletree-deakins-the-mark-of-the-bot-when-employ-3]: **Ogletree Deakins commentary** — "Federal and state laws require employers to provide reasonable accommodations for an employee’s religious beliefs, unless it would impose an undue hardship." *Ogletree Deakins, The Mark of the Bot: When Employees Raise Religious Objections to Workplace AI Usage.* [^microsoft-support-use-copilot-in-word-with-a-scr]: **Microsoft Support, Use Copilot in Word with a screen reader** — "Use Copilot in Microsoft Word with just your keyboard and a screen reader to draft new content, rewrite or summarize text, generate ideas, and ask questions about your document." *Microsoft Support, Use Copilot in Word with a screen reader.* [^microsoft-support-use-copilot-in-outlook-with-a]: **Microsoft Support, Use Copilot in Outlook with a screen reader** — "Microsoft wants to provide the best possible experience for all our customers. If you have a disability or questions related to accessibility, please contact the Microsoft Disability Answer Desk for technical assistance." *Microsoft Support, Use Copilot in Outlook with a screen reader.* [^microsoft-support-basic-tasks-using-a-screen-rea]: **Microsoft Support, Basic tasks using a screen reader with Copilot in PowerPoint** — "This article guides you through basic tasks using Copilot in PowerPoint with your keyboard and a screen reader." *Microsoft Support, Basic tasks using a screen reader with Copilot in PowerPoint.* [^google-accessibility-google-workspace-for-divers]: **Google Accessibility, Google Workspace for diverse learners** — "We create Google for Education products with accessibility in mind, to help every student learn, be inspired, and achieve their full potential." *Google Accessibility, Google Workspace for diverse learners.* [^google-accessibility-help-google-workspace-admin]: **Google Accessibility Help, Google Workspace admin guide to accessibility** — "To make it easier for companies, educational institutions, and government agencies to comply with accessibility standards, we provide transparent information about how our products currently work for people with disabilities." *Google Accessibility Help, Google Workspace admin guide to accessibility.* [^new-jersey-division-on-civil-rights-guidance-on-3]: **New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination** — "precludes or impedes" *New Jersey Division on Civil Rights, Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination.*