Two years after ChatGPT reshaped the conversation about education, most teachers still haven't received formal AI training. The EdWeek Research Center surveyed 1,135 educators in October 2024 and found that 58 percent had gotten no AI professional development at all. For special education teachers, that number carries more weight than it does for anyone else in the building.
A general education teacher without training might turn in a weak lesson plan. A special education teacher without training is writing legally binding documents. IDEA doesn't pause while districts write their AI policies. IEPs still have to be measurable and defensible, and the tools educators are reaching for to write them, faster and under real workload pressure, are increasingly AI-powered.
"I feel that we are at a disadvantage."
High school social studies teacher, Ohio · EdWeek Research Center, October 2024That's a general education teacher talking. Now picture the same gap for a special educator, who also carries legal responsibility for every student's individualized plan.
Teachers are already using it
A Center for Democracy and Technology survey found 57 percent of special education teachers used AI to help develop IEPs or 504 plans in 2024-25, up from 39 percent the year before. Fifteen percent used it to write full IEPs. Thirty-one percent used it to spot trends for goal-setting. Twenty-eight percent used it to help choose accommodations.
Most are doing this without training, district guidance, or any way to check whether the output meets the legal bar an IEP requires: student-specific, data-driven, built with a full team. They're doing it because 45 states reported special education teacher shortages in 2024-25, and these tools feel like the only way to keep up.
What the EdWeek Survey Found
- 1,135 educators surveyed nationally, Sept. 26 to Oct. 8, 2024
- 58% had received no AI professional development
- Only 6% were getting ongoing AI training
- Of those trained, 41% rated it "poor" or "mediocre"
- Lack of knowledge and support was a top reason teachers avoided AI
- Competing priorities, paperwork and compliance for SpEd teachers, was the other
For special education teachers, those two barriers are the same barrier. The paperwork that makes AI tools appealing is the same paperwork where untrained use creates real legal risk.
Training quality matters just as much
Among teachers who did get training, 41 percent called it "poor" or "mediocre." Only 18 percent called it "good" or "excellent." Training is happening. It isn't working.
A general awareness session might be enough to introduce a curriculum tool. It isn't enough to teach a special educator how AI can encode bias, or whether an AI-generated goal actually meets IDEA's "measurable" standard.
"As districts struggle to draft policies and create professional development to train teachers, special educators are already using the tools, with or without guidance."
Education Week, October 2025A University of North Carolina study found special educators trained on AI for IEP goal-writing produced goals rated more favorably than an untrained control group. Training changes outcomes. Without it, the risk lands on students, whose IEPs are the record of what their school promised.
The equity gap
The training gap isn't evenly distributed. RAND's 2024 district survey found low-poverty districts were far more likely to have trained teachers on AI than high-poverty districts: 67 percent versus 39 percent. High-poverty schools also carry the most severe SpEd shortages and the highest caseloads.
The teachers most likely to use AI without training work in the least-resourced schools, with the highest-need students, and the least support. That's an equity problem with a compliance dimension attached.
On legal exposure: The Center for Democracy and Technology notes that AI-generated IEP goals built on limited student data, without significant teacher review, likely don't meet IDEA requirements. Districts that haven't trained staff aren't insulated from that risk. They're exposed to it.
What effective training looks like
Not a 90-minute session on the basics of ChatGPT. Role-specific, ongoing learning built around the real work: IEP development, compliance, progress monitoring, family communication. Clear guidance on what these tools can and can't do under IDEA and Section 504. And it has to repeat. Only 6 percent of teachers were getting ongoing AI training as of October 2024, and the tools change every quarter.
The bottom line
Special education teachers are already using AI, and that isn't changing course. The real decision districts are making, whether they say so or not, is whether that use will be trained and legally grounded, or left improvised and unsupported.
The gap is documented. The risk is documented. What's missing is urgency, and for teachers carrying federal compliance on top of every student on their caseload, urgency is the only reasonable response.
Primary source: EdWeek Research Center survey of 1,135 educators, Sept. 26 to Oct. 8, 2024. Additional data: Center for Democracy and Technology survey of 275 special education teachers, June to August 2025. RAND American School District Panel, fall 2024. University of North Carolina IEP goal-quality study, 2024.