Ethical Considerations of Using AI for Early Autism Prediction

Researchers are developing artificial intelligence (AI)-based prediction tools that use information from electronic health records to help identify children who may be autistic earlier in life. These tools could help families get evaluations and services sooner, but they also raise important ethical questions about fairness, trust, consent, caregiver involvement, how results should be shared, and how results can be used. 

We conducted this study to understand how caregivers of autistic children, autistic adults, and clinicians view the use of AI-based prediction tools for early autism identification tools. Their perspectives can help guide how these tools are developed and used in ways that are ethical, trustworthy, and helpful for families.

The study included 26 adults from the United States: 

• 8 caregivers of young autistic children 
• 8 autistic adults
• 10 clinicians involved in autism screening, diagnosis, or intervention

Participants took part in interviews or small-group discussions held online between March and July 2025. Researchers asked them about their views on autism screening, the use of AI-based autism prediction tools, concerns about ethics and fairness, and how screening results should be communicated to families.

Participants saw potential for AI-based prediction tools to support earlier autism identification by helping detect patterns in health records that traditional screening methods may miss. This could be helpful generally but may be especially helpful for groups for whom autism is more often missed. 

However, participants raised concerns about bias, gaps in health records, lack of transparency about how AI makes predictions, and the possibility of increasing stress for families. Most participants preferred using AI predictions together with caregiver-reported information rather than relying on either source alone. 

They also emphasized the importance of clear communication, clinician guidance, and providing families with practical next steps after receiving results. Participants viewed earlier identification as most useful when the information could be acted upon. Telling a family about an increased likelihood for autism before a formal diagnosis is possible can limit the family’s options for next steps.

The findings suggest that people see potential benefits in using AI-based prediction tools to support earlier autism identification, but successful implementation will require more than accurate technology. Families, autistic adults, and clinicians need tools that are fair, transparent, and easy to understand. They also want caregivers to remain involved in the process and to receive clear explanations and actionable recommendations. These results highlight the need to build ethical practices into the design and use of AI-based prediction tools so they can support trust, equity, and shared decision-making when used in autism clinical care.

Read the full open access paper at:

Sperling, J., Haseley, E., Quenstedt, S. et al. AI–based clinical prediction models for early autism detection: ethical considerations and use implications across caregivers, autistic adults, and clinical providers. BMC Med Ethics (2026). https://doi.org/10.1186/s12910-026-01634-w