In special education, AI tools such as text-to-speech, speech-to-text, and adaptive reading software are treated as assistive technology and fall under existing rules: the Individuals with Disabilities Education Act (IDEA) and the Individualized Education Program (IEP) process in the United States, and IEP-based funding rules such as Ontario's Specialized Equipment Allocation in Canada. Early research is promising but thin. These tools support teaching and do not replace an evaluation by qualified professionals.
Key takeaways
- Under IDEA regulations, any item that increases, maintains, or improves a child's functional capabilities can count as an assistive technology device, whether or not it uses AI.
- IEP teams in the United States must consider whether a child needs assistive technology, and Ontario's 2026-27 equipment directives tie funding to needs documented in an IEP.
- A 2018 meta-analysis found a modest benefit from text-to-speech tools, and a 2025 systematic review found positive but methodologically weak evidence for AI-based interventions.
- Student data privacy, accessibility, and over-reliance need review before any tool reaches a classroom.
How IDEA and the IEP frame AI tools
Federal regulations define an assistive technology device as any item, piece of equipment, or product system, bought off the shelf, modified, or customized, that is used to increase, maintain, or improve the functional capabilities of a child with a disability. A speech-to-text app qualifies by function, not by label. Section 300.105 requires each public agency to make such devices or services available when they are required as part of special education, related services, or supplementary aids and services. Section 300.324 adds that the IEP team must consider whether the child needs assistive technology devices and services.
The definition of assistive technology service also includes evaluating a child's needs, including a functional evaluation in the child's customary environment. In practice, a district cannot adopt an AI reading tool for a whole class and call it an accommodation. The decision belongs in the IEP process for an individual child.
Canadian rules: the Ontario example
Canadian provinces set inclusive education policy separately, so no single national rule exists. Ontario's Ministry of Education publishes Specialized Equipment Allocation (SEA) directives, and the 2026-27 edition (Spring 2026) funds equipment that a student needs to access the Ontario curriculum, meets a physical, personal care, auditory, communication, or visual need, and is outlined in the student's IEP. Eligible software includes reading and math intervention software that meets those conditions. Leasing and service contract costs are covered under CAD 5,000 per year before taxes.
A professional assessment is not required for an IEP, but the claims-based part of SEA funding always requires one, and boards are encouraged to seek one for safe and appropriate use of equipment bought with formula funding. Boards must also tell parents or guardians when a claim that shares student information has been submitted. Rules in other provinces were not reviewed for this article.
Tool types and what to check
The table groups common AI-related tools by purpose. The checks are suggested questions for an IEP team, not legal requirements.
| Tool type | Typical purpose | What to check before adoption |
|---|---|---|
| Text-to-speech and read-aloud | Access to written text for students with reading difficulties | Voice quality, ability to follow along on screen, and fit with the exact reading barrier |
| Speech-to-text | Writing support for students with motor or writing difficulties | Accuracy with the student's own voice and accent, and whether recordings leave the device |
| Adaptive reading and math platforms | Practice that adjusts to a student's responses | How difficulty is decided, whether teachers can override it, and what data is stored |
| Generative chatbots and writing aids | Drafting, summarizing, explaining | Age limits, accuracy of outputs, and whether use replaces the skill being taught |
What research shows so far
The strongest evidence concerns older, narrower technology. A 2018 meta-analysis in the Journal of Learning Disabilities (Wood, Moxley, Tighe, and Wagner) examined text-to-speech and related read-aloud tools for students with reading difficulties. It reported an average effect size of 0.35, with a 95 percent confidence interval of 0.14 to 0.56, and called for more studies on what shapes the effect.
For AI-based interventions, a 2025 systematic review in Brain Sciences (Paglialunga and Melogno) covered 11 studies with 3,033 participants, six of them on dyslexia. All 11 reported positive outcomes, yet none was rated low risk of bias. The authors asked for randomized controlled trials and longer follow-up, and they named cognitive offloading, where a tool does the thinking, as a risk to watch. Those results justify cautious trials, not broad claims.
Privacy, accessibility, and policy signals
The Family Educational Rights and Privacy Act (FERPA) regulations apply to educational agencies and institutions that receive funds under programs the US Secretary of Education administers. Any AI vendor handling student records therefore raises a compliance question for any school or district that receives such funds. In the final priority on advancing AI in education (91 FR 18774, published April 13, 2026, effective May 13, 2026), the Department of Education noted its July 2025 grant guidance recommending that AI tools be accessible to users with disabilities, and it added a paragraph supporting universal design for learning. The priority applies to discretionary grant competitions and is not a regulation on schools.
Universal design for learning (UDL), from CAST, organizes planning around engagement, representation, and action and expression. It treats flexible options as the default, which is a more durable starting point than any single AI product.
A pre-adoption checklist
- The tool addresses a barrier named in the child's evaluation or IEP.
- A qualified professional has contributed to the assessment of need.
- The vendor's data practices are written down, including storage, retention, and training on student content.
- The tool works with the student's other accommodations and devices.
- Progress is measured against a stated goal and reviewed at set intervals.
- Families receive a plain explanation of what the tool does.
AI tools do not diagnose disabilities and do not replace professional assessment. IDEA's evaluation rules call for a variety of assessment tools, bar any single measure as the sole criterion, and require trained and knowledgeable personnel to administer assessments. Families with questions about a child's needs can start with the school's special education team or a licensed clinician.
FAQ
Is an AI app automatically assistive technology?
No. Under the federal definition, function decides. An app counts as an assistive technology device when it is used to increase, maintain, or improve a child's functional capabilities.
Can a school require an AI tool without an IEP decision?
A school can use tools for a whole class as general teaching practice. When a tool serves as an accommodation or service for a child with a disability, the IEP team is the body that decides.
Does the research prove AI helps students with disabilities?
Not yet. Reviews report positive results, but the studies are few and carry a moderate or serious risk of bias. The text-to-speech evidence is older and somewhat stronger, with a modest average effect.
Where does Canadian policy stand?
Each province sets its own policy. Ontario's equipment funding rules link support to IEP-documented needs and, for claims, to a qualified professional's assessment.
Related reading
- AI and inclusion in special education
- current trends in AI in education
- how AI technology affects schools
- potential of AI in education
More in AI in education.
Last verified: October 2026
Sources
- 34 CFR 300.5, 300.6, 300.105, 300.304, and 300.324, Electronic Code of Federal Regulations, US Government
- 34 CFR 99.1, Family Educational Rights and Privacy Act regulations, Electronic Code of Federal Regulations, US Government
- Final Priority and Definitions, Advancing Artificial Intelligence in Education, 91 FR 18774, Federal Register, April 13, 2026
- Specialized Equipment Allocation (SEA) 2026-27 Directives, Ontario Ministry of Education, Spring 2026
- Wood, S. G., Moxley, J. H., Tighe, E. L., and Wagner, R. K., Does Use of Text-to-Speech and Related Read-Aloud Tools Improve Reading Comprehension for Students With Reading Disabilities? A Meta-Analysis, Journal of Learning Disabilities, 2018
- Paglialunga, A., and Melogno, S., The Effectiveness of Artificial Intelligence-Based Interventions for Students with Learning Disabilities: A Systematic Review, Brain Sciences, 2025
- Universal Design for Learning, CAST