A working paper from the OECD examines the opportunities, risks and limitations of AI being used to help students with special educational needs (SEN) achieve their learning goals.
Last month (September 2025), the OECD published a new working paper in their OECD Artificial Intelligence Paper series, Leveraging Artificial Intelligence to Support Students with Special Education Needs.
The paper uses research-based evidence to explore the current landscape of AI tools for students with SEN, the risks and limitations of AI tools for students with SEN, effective operations and governance, and policy considerations.
Key findings
- There are some promising examples of AI tools that might respond to the specific needs of students under the umbrella of SEN.
- There is a lack of data on how AI tools may be used systematically by teachers and students with SEN.
- Data privacy is a real ethical concern for using AI with students, especially those with SEN. For example biometric data, e.g. eye gaze and facial recognition could be misused or commercialised.
- Inaccurate and non-evidence based material online poses a risk. This material can be incorporated into LLMs, meaning recommendations and outputs for children with SEN may have issues with accuracy and appropriateness.
- Similarly, reinforced biases, lack of accountability for harm, perpetuating inequalities and discriminatory practices were large concerns – these can lead to inaccurate predictions about needs or diagnoses, stereotyping and underrepresentation of groups, false positive with facial recognition etc.
- Positive initiatives from the health field can be beneficial, such as the development of a certification for the transparent use of AI, protection of rights of students with SEN in research and open data standards.
- Co-design and international collaboration have been recognised as crucial in designing, developing and implementing AI systems that are safe, ethical and beneficial.
- In the UK, some teachers reported generative AI supported students with SEN by increasing student engagement and facilitating the adaptation of the learning content – such as developing worksheets that students could use to progress at their own pace.
- In Norway, teachers recognised AI helps teachers and students explore different genres and create various artefacts, such as advertisements and informative material in a marketing class.
- In a survey in the USA, 51% of teachers in K-12 classrooms reported using AI tools to support students with learning differences.
- A common concern amongst teachers in Norway was students’ ability to think independently, comprehend and apply knowledge without relying on technology (such as AI chatbots).
- UK parents and students raised concerns over access to AI tools, that it would be limited to schools who could afford it and exacerbate inequalities in education, the labour market and society. 1
Reference
- OECD – Leveraging Artificial Intelligence to Support Students with Special Educational Needs (2025). Available at: https://www.oecd.org/content/dam/oecd/en/publications/reports/2025/09/leveraging-artificial-intelligence-to-support-students-with-special-education-needs_ebc80fc8/1e3dffa9-en.pdf (Accessed 9th October 2025)
