Home $ Blog $ Ten Things about Learners with Disabilities in the Age of AI
Back to All Posts

Ten Things about Learners with Disabilities in the Age of AI

DATE

September 18, 2026

SHARE

Artificial intelligence (AI) is reshaping the educational landscape at an unprecedented pace, presenting both opportunities and risk for K-12 classrooms. While AI tools hold the potential to reduce barriers and broaden access, the use of AI in special education requires human oversight that is deliberate and ethical. Legal mandates such as the Individuals with Disabilities Education Act (IDEA) and Section 504 of the Rehabilitation Act of 1973 remain at the forefront, meaning governance, data privacy, and human-driven decision making cannot be outsourced. Designed for educators (general and special education) and school leaders, this blog post outlines ten guiding principles to navigate AI adoption responsibly, ensuring the technology empowers and supports learners with disabilities through best practices in teaching and learning.

In the Classroom

1. Level up your own AI literacy

We cannot guide students through something we do not understand ourselves. That is true with everything, including with AI tools. AI readiness means understanding how AI works, where it helps, and where it can fail. AI literacy goes further in considering how to use AI to solve real problems and improve practice (Florida K-12 AI Education Task Force, 2025). For those of us working with learners with disabilities, a third layer matters just as much:

  • Knowing what the assistive technologies our students already use actually do, and how AI has changed them
  • Being able to recognize when an AI tool creates an access barrier instead of removing one
  • Understanding enough about data privacy to know what can and cannot go into a tool
  • Recognizing our own expertise as the thing that makes AI useful here. AI does not know our students. We do.

Practical ways to build our own AI literacy includes:

  • Try a district-approved AI tool yourself before using it with students.
  • Learn alongside colleagues rather than alone. Working with grade-level teams and special education/exceptional education departments can help
  • Talk with your assistive technology specialist. They have been evaluating tools against individual student needs for years, and that skill transfers directly

Strategy: Each month, pick one approved AI-enabled tool you are curious about and spend twenty minutes exploring it. Think about the needs of your learners and consider how this tool either reduces or creates barriers for your learners. Over time, create a list of these tools. Also, take advantage of AI literacy professional development offered through your school district or through state and national initiatives.

2. Start with the student, not the tools

Before deciding on an AI tool, ask three questions:

  • What do we want the students to know and be able to do?
  • What are the barriers to the student(s) full participation?
  • Could AI help address this specific barrier in ways that are not currently being addressed?

Answering these questions aligns with the Universal Design for Learning (UDL) framework, which asks us to reduce barriers in advance and offer more than one way of engaging learners, representing information, and allowing learners to express their understanding (CAST, 2018). AI has the potential to strengthen UDL when we use it to widen access.

We should consider using AI when it serves a clear instructional purpose: Helping a student understand directions, organize ideas, communicate a response, or access information they otherwise could not reach. We should avoid adopting a tool just because it is new, is free, or is being used down the hall by another teacher.

Strategy: Name the barrier before you name the tool. If you cannot state the barrier in one sentence, you are not ready to choose a tool.

3. AI Assists. People Decide.

AI may help us as educators brainstorm strategies, organize information, or draft preliminary language. It must not independently determine a student’s special education eligibility, academic and behavioral goals, accommodations, services, instructional placement, or progress. Educators, families, specialists, and, whenever possible, the students themselves, should make instructional decisions using reliable information about that student.

Strategy: Never put a student’s name into an AI tool. Instead, take the draft language the district-approved tool gives you, and consider whether it addresses the individualized needs of the student. If the draft language could describe any student on your caseload or in your classroom, it is not individualized.

4. Keep expectations high and reduce barriers

For some students with disabilities, understanding new or complex information can be exhausting. If the cognitive load is too high, they may not have the chance to authentically engage with the content in meaningful ways. Addressing this barrier means that their energy can be spent on deeper learning. AI has the potential for reducing cognitive load. AI tools can help create:

  • Chunked directions
  • Plain-language explanations
  • Vocabulary supports
  • A range of worked examples
  • Visual supports
  • Practice questions

The important thing to remember is that we should not reduce the learning expectations when we offer AI-enabled scaffolding or supports. Making information easier to access helps students engage with content without reducing the expectations or rigor of what they are expected to learn. There is a difference between removing a barrier and removing the work expectation. Productive struggle is part of learning, and a scaffold that does the thinking for a student has resulted in over-accommodation rather than support.

Strategy: Before offering an AI scaffold, name the goal and/or skill in the lesson. Then confirm the scaffold does not alter the expectations.

5. Double check for accuracy before you share

AI can produce material quickly, and this quick output does not necessarily translate to accuracy. Wrong information can be beautifully formatted, plausible, and appear ready for sharing. However, this output can be inaccurate, confusing, or not individualized for your students. Additionally, when we ask AI to simplify content, it may strip critical information along with the hard vocabulary. Additionally, we need to pay attention to auto-generated captions and alt text. Although these are getting more and more accurate, they can be inaccurate, especially with technical or content-specific material. Never publish auto-generated alt text or captions without reading them against the source.

The important thing to remember is that what is generated by AI may be easier to read or access, but it may no longer reflect what we want to teach.

Before students use AI-generated materials, check:

  • Is the information accurate?
  • Was rigor reduced, important ideas removed, or changed in the simplification?
  • Is the language age appropriate?
  • Did simplification remove the ideas that mattered?
  • Are images, captions, alt text, descriptions, and examples accurate?

Strategy: Build the accuracy and accessibility check into your lesson planning process from the very beginning. Read the AI produced content against the original sources before you share.

6. Bring your students into the conversation

Students with disabilities are the people most affected by decisions about AI tools and the ones that are typically least often asked. However, listening to students will often tell us what a long procurement review may take months to discover.

The Individuals with Disabilities Education Act (IDEA) already requires that, when possible, students be invited to their own IEP meetings, especially once postsecondary goals and transition services are discussed. Including learners in these decisions helps build their agency and independence. If a student cannot attend these meetings, we still must make sure that their preferences and interests are considered

What this looks like in practice:

  • Ask students directly whether a tool helps. If a student states that the tool is not helpful, that information is critical as they will not use tools that are not effective for them.
  • Teach students to use their AI supports rather than just handing them over.
  • Help students explain their own support needs. Encourage them to state their needs and supports (e.g., “I use text-to-speech because long passages are slow for me. I can also check the summary against the original”).
  • Help students reflect on how their independence may change with the use of AI. This focus on student agency will outlast any particular tool. As AI becomes more common in schools, independence increasingly means knowing when, why, and how to use available supports well, not doing every task unaided. Teaching students to make that call is part of self-advocacy.
  • Explicitly talk about AI use disclosure with students. AI disclosure is a decision with real social consequences, so it is better taught than left to chance.
  • Include students in tool selection where you can. A student panel comparing two options will find problems no adult noticed.

Strategy: Ask each student who uses an AI-supported accommodation these questions: Does the tool help? If so, under what conditions? Write down what they say and bring it to the next IEP or 504 meeting.

7. Ditch the cheating detectors

AI writing detectors flag some learners’ writing more than others, and students with disabilities are among those most exposed to these types of errors. Currently, AI-writing detectors are not reliable enough to act on for disciplinary action. A peer-reviewed evaluation of 14 tools, including Turnitin, concluded that they had issues with reliability and accuracy (Weber-Wulff et al., 2023).

AI detectors score text partly on lexical variety and predictability, so any writer whose writing is different from the expected norms may be inaccurately flagged as cheating. For example, Liang et al. (2023) found that detectors classified 61% of essays by non-native English writers as AI-generated, while performing near-perfectly on U.S. eighth-grade essays. The same mechanism puts students with disabilities, those using grammar assistants, or translation tools at risk for being flagged for using their supports.Thus, AI detector scores should never be a sole basis for accusing or disciplining a student.

Strategy: Set clear expectations up front for when and how students may use AI. Most integrity disputes are really expectation disputes. Also, consider multiple forms of evidence including:

  • Drafts
  • Revision histories
  • Notes and outlines
  • Citations
  • Conversation with the student
  • Other demonstrations of understanding

Schoolwide Responsibilities

8. Guard student needs and privacy

Before using any AI-enabled tool, we need to think carefully about whether these tools are allowed, appropriate, and protect student data. Student education records remain under district control, and vendors performing institutional services are bound by the same obligations (FERPA: Family Educational Rights and Privacy Act, 1974). Districts have strict rules about these, so it is important to know these rules. Some of the common rules include:

  • Only use AI tools that have been reviewed and approved through the district’s technology adoption processes.
  • Staff should not enter student names, disability information, IEP or 504 content, behavioral notes, identifiable work samples, photographs, recordings, or other private information into AI tools unless that specific use has been approved and appropriate protections are in place.

Strategy: Before adopting a tool, teachers and administrators should establish:

  • What information the tool collects
  • Where the information is stored
  • Who can access this information
  • How the information is used
  • How long it is retained
  • Whether it is used to train an AI system
  • How families and students will be informed, including why AI is being used, how it supports learning, what safeguards are in place, and how educators (not the technology) continue to make instructional decisions

9. Disability laws still apply

Using AI-enabled tools does not change our obligations under the Individuals with Disabilities Education Act (IDEA, 2004), Section 504 of the Rehabilitation Act (1973), or the Americans with Disabilities Act (ADA, 1990). There is no AI exemption and no separate set of rules for use of AI-enabled tools. Schools remain responsible for making sure students with disabilities can access and participate in educational programs, even when an outside company provides the technology. Put simply, buying and using software does not transfer a FAPE obligation to a vendor. If a tool creates a barrier for a student, the answer cannot be that the vendor built it that way.

Two areas deserve particular attention, because AI changes the risk rather than simply inheriting it.

AI repeats the patterns in its training data. AI systems learn from historical data and can carry those patterns forward. This fact matters most in high-stakes areas such as disability determinations, instructional placement, discipline, and grading. If a tool is informing any of those decisions, someone needs to be watching its outputs by student population, not just for overall accuracy.

AI-informed decisions still have to be explainable. When AI contributes to a decision affecting a student’s education, we need to be able to explain in plain language what the tool does, what information it uses, and how its output shaped the outcome.

Strategy: Before adopting an AI tool, ask the question: If this tool fails a student with an IEP next month, what is our answer? If the answer is that we would ask the vendor, we have not yet accepted the obligation that IDEA, Section 504, and the ADA place on us.

10. Demand accessible AI-enabled technologies

We are the buyers of learning tools, and accessibility belongs in the selection process, not in a conversation after the contract is signed. The Americans with Disabilities Act (ADA) also carries an accessibility deadline. Under the U.S. Department of Justice’s Title II rule, technologies such as websites, mobile apps, and digital materials must meet WCAG 2.1 Level AA accessibility compliance by April 26, 2027 for entities serving 50,000 or more people, and by April 26, 2028 for smaller entities and special district governments. This requirement includes AI-enabled learning platforms, digital curriculum, and online assessments, which means purchasing decisions made this year already carry a compliance date.

Vendor claims about accessibility are a starting point, not proof of accessibility compliance. Many companies will provide an Accessibility Conformance Report, usually based on the Voluntary Product Accessibility Template (VPAT). These are useful for comparing products against a specific feature, but they are self-reported, and their quality depends entirely on the expertise of the people who complete them. Thus, these reports tell us what a vendor believes about their product. It does not tell us whether our students can use it.

Before purchasing AI-enabled technologies, ask vendors to demonstrate that students can complete the essential tasks using:

  • A keyboard, with no mouse or trackpad
  • Screen readers and text-to-speech tools
  • Speech input and alternative input devices
  • Captions and transcripts
  • Magnification and enlarged text
  • High-contrast and customized displays

Strategy: Run a five-minute keyboard test on any tool under consideration. Can you enter a prompt, read the response, and save your work without touching the mouse? Then find out who in your district owns digital accessibility. If no one can answer that question, that is the first problem to solve.

Special thanks to Daniela Simic, Thomas Kennedy, & Roberto Alonso for the contributions and feedback.

Resources

Translate