New Mexico's Public Education Department mandated the use of Amira Learning, an AI-based reading assessment that records students reading aloud, as part of the state's K-3 literacy screening under the Structured Literacy Act. Santa Fe Public Schools paused the software in late 2023, citing concerns about how student voice recordings are stored, shared, and retained.
District concerns over voice data
Amira Learning's software uses artificial intelligence to listen to students read aloud and score oral reading fluency. The tool records audio of children's voices, which constitutes biometric data under some privacy frameworks, according to the company's privacy policy (source 2). The Santa Fe superintendent communicated to families that the district needed to ensure adequate data privacy protections and parental notification before resuming the software, as reported by the Santa Fe New Mexican (source 1). The school board discussed the pause in a public meeting, with members citing the need for clearer consent processes before recording young children's voices, according to board minutes (source 4).
State mandate and privacy rules
The New Mexico Structured Literacy Act, House Bill 297, requires K-3 students to be screened for reading difficulties using state-approved instruments. The state education department designated Amira as one of those tools (source 3). Amira Learning's published privacy policy states that student data including audio recordings is used to provide the service and improve the AI model, and that the company complies with the Family Educational Rights and Privacy Act, the Children's Online Privacy Protection Act, and state student data privacy laws (source 2). Santa Fe officials questioned whether those protections were sufficient, particularly for kindergarten students.
South Carolina precedent
South Carolina implemented Amira statewide for K-3 reading assessment through its Read to Succeed Act. The state encountered parent opt-out requests and questions about voice data. The South Carolina Department of Education issued guidance clarifying that parents could not opt out of required state assessments but could request information on data practices (source 5).
Research on AI reading assessment
A peer-reviewed study published in the Journal of Educational Psychology found that AI-based oral reading fluency assessment tools like Amira showed strong correlation with human-administered DIBELS assessments, with correlation coefficients above 0.85 for fluency scoring. The study noted limitations including potential bias in speech recognition accuracy for non-standard English dialects (source 6).
Another study, published by the American Educational Research Association, found mixed results on whether AI-based reading tools improve reading outcomes compared to traditional teacher-administered assessments. While the tools provided more frequent assessment data, the study found no statistically significant improvement in end-of-year reading scores for students using AI tools versus those assessed by teachers alone (source 7).