What the Legislature was told about Amira.
Near the end of a July interim hearing in Taos, the chairman of the Legislative Education Study Committee described his own son — an excellent reader with an IEP for a speech impediment — who “gets absolutely frustrated and really dislikes” the state’s AI reading assessment because it scores him poorly. Then he asked the vendor’s chief executive what the company does with student data, and whether it ever deletes it.
The answers — and the contract structure, the rebid timeline, a speech-content alert system, and the retention rationale that came out in the same ninety minutes — are now part of the public record. The next afternoon, the public answered back: six commenters, from Santa Fe parents to a Roswell district administrator, put their own experience on the same record. This page holds both days, timestamped. Every quote links into the archived webcast at the moment it was said.
Quotes were transcribed from the archived legislative webcast by machine transcription, then each quoted passage was independently re-verified against the audio with a second, larger transcription model. Timestamps deep-link into the official player so any quote can be checked against the original in one click. If you find an error against the recording, use the correction contact at the bottom of this page; corrections are applied and dated.
Watch the full Amira item in the archived webcast ↗§01The hearing at a glance
The recording runs 3:10:00, beginning at 1:36:03 PM MDT. The Amira item is the final agenda item and fills the last ninety-six minutes.
§02Six things that entered the record
Each entry states the fact as it was put on the record, then — separately — what the fact changes procedurally. The distinction is deliberate: the facts are the hearing’s; the significance lines are this campaign’s reading, and are labeled as such.
Asked about the current Amira contract, PED's assessment director described the procurement structure: the Department does not hold a direct contract with Amira Learning. It piggybacks on an umbrella contract held by Cooperative Educational Services, through a year-to-year intergovernmental agreement — “we do a year-to-year inter-governmental agreement between us and Cooperative Educational Services to acquire Amira.”
why it mattersThis is the first public explanation of why the statewide contract's value and terms have been hard to locate in the Sunshine Portal. The records now have a precise address: the CES umbrella contract and the annual intergovernmental agreement are both public records. The IPRA template on this site has been updated to name them.
The replacement RFP is “fully developed,” with one piece being added. PED plans to route it to the General Services Division within about three weeks, with public notification estimated in the next three to four months. The award is intended for transition and professional-learning planning during 2026-27, “where the year that it would be in front of students” is the 2027-28 school year. Vasquez also stated the procurement exists in part “to be able to better examine the types of screeners that are out there and for us to vet.”
why it mattersThe procurement clock this campaign's asks run against now has dates from the Department itself — and PED's own framing of the rebid is a re-vetting. Every vetting, privacy, and accommodation standard a district or the Legislature wants in place has a vehicle: the RFP that goes public this fall.
Unprompted, PED's assessment director described a feature she said “people probably don't even know” — a “troubled child alert” developed with Amira. If a student speaking to the system says something “irregular” — her examples were “I didn't get any breakfast this morning” and “I don't feel safe at home” — an alert goes to the district test coordinator. She called it “a safety net to protect students.”
why it mattersThis is the clearest official statement to date that the system analyzes the content of what children say into the microphone, beyond scoring how they read. Whatever one concludes about the feature — and child-safety alerting is a defensible purpose — its existence, criteria, recipients, and the retention treatment of flagged speech are exactly the kind of facts that belong in a published governance framework and in parent-facing disclosure. Neither currently names it.
The vendor's CEO gave the design rationale for keeping student voice recordings: “this is why we keep a digital recording — teachers can override, district administrators can override, the state can override.” On deletion, later in the hearing: some districts “zap the data on a nightly basis,” in parts of Idaho “we don't store any data at all,” and “most places, we purge the data at the end of the school year… we'll hold it or delete it at your discretion.”
why it mattersRetention is a configuration, chosen by the education agency — not a technical necessity. That makes New Mexico's actual configuration a specific, answerable public-records question: what did the state choose? The IPRA template on this site now asks.
In 2023-24, 45% of screened New Mexico students were flagged as possibly at risk for dyslexia — against national prevalence estimates of 10-20%. Last year the statewide rate fell to 23.3%; the cohort screened with Amira flagged at 16.7%. Under questioning about the earlier rate, Vasquez named possible factors including “teacher bias” and differing technical expectations across screeners, and said it is “time for us to reflect on our policy.”
why it mattersA screening system that flagged nearly half the state's children, and a correction this large, is a measurement-governance story in either direction — and it is the strongest argument in the record for the vetting and evaluation standards the Legislature's own analyst found missing in May. The numbers cut both ways and are presented here as PED presented them.
PED confirmed the accommodations scoping: a paper form of the assessment is available upon request — “which means it has to be written into an IEP.” A non-verbal, keyboard-only version also exists as an accommodation. The vendor's CEO later added that a Braille version exists and that alternatives are available for students who cannot work well “in the classic verbal mode.”
why it mattersFor a parent deciding what to request in writing before the first assessment window, this is the operative constraint: the alternative formats exist, and the district-level question is who can access them and through what process. A parent without an IEP for their child currently has no stated route to the paper form.
§03In their words
Verbatim, verified against the recording, in context. The context lines are this page’s; the words are theirs.
“Some screen time is literally cocaine — it's literally designed to do all the wrong things for our kids. And some screen time is chemotherapy — it's life-saving.”
Responding to a screen-time question; Angel noted he had said the same on CNN. The vendor's own frame concedes the premise that screen time in classrooms requires justification — the question the campaign's asks put to the district is who verifies which category a given tool is in, and how.
“Amira's a huge advocate for regulation around ed tech. We believe it needs to be far more regulated than it is today.”
Followed by a description of outcomes-based contracting — compensation contingent on measured student growth — which Angel said Amira has done with Indiana, Michigan, and large Florida districts. A vendor stating on the record that it supports stronger regulation, and that it accepts outcome-contingent contracts elsewhere, is directly relevant to what New Mexico can ask for in the upcoming rebid.
“Teachers often reach conclusions that, I'll just say frankly, the data disputes.”
The close of Angel's answer to a legislator's account of teachers reporting that the system misreads accented students. The exchange — teacher observation versus system output, and which is treated as ground truth — ran through the entire Q&A.
“My son is taking Amira, and he's a really excellent reader — has been for a long time — but has an IEP for a speech impediment. He gets absolutely frustrated and really dislikes taking Amira because of his results, which tend to be very poor.”
The committee chair, describing his own child's experience before asking the vendor directly what it does with student data, whether it deletes data, and what purposes the retained data serves. The privacy questions in this hearing came from the chair.
“Most places, we purge the data at the end of the school year… but we'll hold it or delete it at your discretion.”
Answering the chair's question, “Does your company ever delete data?” The answer places New Mexico's retention configuration squarely in the state's hands — and makes it a question the state can be asked in writing.
“Over the course of this past school year, Amira conducted almost 600,000 discrete assessments in New Mexico… each assessment involves usually multiple thousands of measurement points.”
The vendor's own description of collection scale, offered as a strength. Taken with the roughly 50,000 students in the K-2 mandate plus grade-3 uptake, it is the clearest single number for how much measurement of New Mexico children now flows through this one system per year.
§04What the committee pressed on
The questions came from members of both parties, and several came from direct classroom or family experience. Members are named only where the recording identifies the speaker with confidence; the full roster is on the agenda.
Longitudinal data access — whether a student's results follow them across grades and districts, and who decides; the 45% dyslexia flag rate and whether a root-cause analysis was done; Amira's predictive validity against summative assessments.
Field reality versus vendor claims: teachers reporting that the system does not recognize students' accents, proficient students routed into remediation, teachers running duplicate assessments to trust the results, Amira passages being printed for practice, and the supervision burden of the required 30 minutes. Asked when the RFP opens and argued districts need choices among assessments.
Accent recognition — “to pretend that I don't have the knowledge because I speak different than other people, that's a disservice to our students” — plus cultural bias in passage content and cumulative screen time across programs.
Students with speech impediments, from direct family experience; what data teachers actually receive; who owns the data; whether the vendor deletes data and what any retained data is used for.
The Legislature’s own May policy brief found the state has no statewide procurement standards for AI tools, no AI-specific student-privacy requirements, and no formal evaluation process for AI education tools. This hearing is what oversight looks like while that gap is open: individual members asking individual questions, with no staff analysis attached to the item.
§05The paper record — what the handouts add
Committee members received a printed packet at the hearing: PED’s fifteen-slide deck followed by Amira Learning’s seventeen-slide deck. A photographed copy of the full packet is archived on this site. The print record confirms parts of the spoken record, adds figures the audio could not carry, and in a few places diverges from what was said aloud. Each entry cites its page in the packet; the significance lines are this campaign’s reading, and are labeled as such.
Amira's state-by-state table lists each state's adoption tier: Texas “One of 3 Approved Screeners,” Oklahoma and Georgia “Only Funded Screener,” California “One of 3 Approved, Only Unanimous” — and New Mexico and Idaho “Mandated Screener.”
why it mattersThat the deployment is a statewide requirement rather than a funded option is the vendor's own printed classification, not a campaign characterization. The mandate is PED administrative policy (the PED deck's own slide: “Required for all K–2 students”), not statute — which is why the upcoming RFP is the vehicle for changing its terms.
About 146 districts and charter schools served; about 53,000 K-2 students took universal and dyslexia screening; 55,487,275 minutes of tutoring; about 574,000 “Progress Monitors (all forms of assessments)” across the state.
why it mattersThe printed counterpart of the CEO's spoken “almost 600,000 discrete assessments.” The printed figure is ~574,000 and is labeled progress monitors; the 55.5 million tutoring minutes is the single clearest published measure of how much recorded student speech the system now processes in New Mexico each year.
The effect-size slide — carrying the NMPED logo — reports New Mexico student counts by grade: kindergarten 13,088; grade 1 15,959; grade 2 15,978; grade 3 10,426; grade 4 8,395; grade 5 7,212; grade 6 2,055. All effects are computed on the Amira Reading Measure (ARM), the vendor's own instrument.
why it mattersRoughly 17,600 students in grades 4-6 used the system last year — well beyond the K-2 mandate and the K-3 the state pays for. The efficacy claims are also self-referential: growth is measured on the vendor's own metric, not on an independent assessment.
A slide titled “Amira Assessment Is Delivering Greater Parity” prints 2025-vs-2026 proficiency: non-ELL 32.7% to 40.1%, ELL 14.7% to 24.6%, parity ratio 0.45 to 0.61.
why it mattersAt the same hearing, PED told the committee its preliminary 2026 results were finalized that morning and that it was “not at liberty” to discuss them. The packet does not say which 2026 window the vendor's figures describe. Whether the vendor's printed 2026 numbers are the results the state declined to discuss is a question the record does not answer — and one the committee, or a records request, can.
The “+7 Weeks Growth in Louisiana” slide reproduces the study cover: “Amira Learning in Louisiana Public Schools, ESSA Level II Study (2023-24) — prepared for: Amira Learning; prepared by: Instructure” (The Edtech Collective), March 2025.
why it mattersThe headline out-of-state efficacy claim rests on a study commissioned by the vendor and produced by a commercial partner, at ESSA evidence Level II (moderate, correlational — not a randomized trial). That provenance was not stated aloud in the hearing; it is on the printed cover.
The MOY regional-training survey (“What challenges have you encountered when using Amira in your classroom?”, 113 responses, multiple selections): technology or device access 55 (48.7%); understanding reports 45 (39.8%); time and scheduling 42 (37.2%); student engagement 36 (31.9%); other 14 (12.4%); no major concerns 9 (8.0%).
why it mattersThe state's own printed data shows 92% of surveyed educators reporting at least one implementation challenge — the quantitative backdrop to the classroom accounts given in committee questions on day 1 and public comment on day 2.
Printed: “A paper form of Amira is available upon request as an accommodation. The paper form still requires students to verbalize on a specific subtest.” Also printed: a non-verbal keyboard version, and that the Assessment Bureau, Amira, and the Office of Special Education are refining accommodations documentation for 2026-27.
why it mattersThe printed slide says “upon request”; the spoken testimony added the operative constraint — “which means it has to be written into an IEP.” A parent reading the handout would not learn the IEP requirement. The 2026-27 documentation revision is the open vehicle for changing who can access the alternative formats.
§06Day 2 — the public answers back
The next afternoon, at the close of the July 16 session, the committee took public comment — up to two minutes per speaker, in person and online. Six of the twelve comments answered the Amira presentation directly: two Santa Fe parents, a grandparent, an Albuquerque parent speaking for a second organized coalition, a district administrator from Roswell, and a retired teacher. The other six covered CTE staffing, paid parental leave, attendance policy, graduation re-engagement, tax policy, and STEM funding — the selection below is by topic, not by viewpoint. No speaker that afternoon spoke in the deployment’s defense.
These are members of the public speaking in a public proceeding. Their words are quoted verbatim and verified against the recording; their characterizations are their own. This campaign documents the record and neither directs nor speaks for any commenter.
“For what reason does he need a tech company or an AI bot to assess his reading? … I propose the following solutions for this upcoming school year: no one-to-one devices in pre-K through fifth grade, a two-year moratorium on AI, computer labs for tech education starting in fourth grade, pencil and paper, period, an opt-in program for voice recording, purposeful tech use that has a daily time limit.”
The first named ask on this record for opt-in — consent obtained before recording — rather than an opt-out exercised after deployment.
“This screen program is only advertised to be used 30 minutes a day. From listening to my grandchild, however, he has access to use it for games and other programs as well. He is six years old. He started this program when he was five.”
A family's account that device use extends beyond the assessed 30 minutes — a factual claim that district device-configuration records could confirm or refute.
“If it had been [made to benefit children], parents and guardians would have been properly notified and asked specifically for consent to have their children's voices recorded. Options would have been presented to include a sincere opt-out, one that doesn't penalize schools. Data privacy standards for not only the vendor, but also PED and the districts would be in place.”
A second organized parent coalition — Albuquerque-based, distinct from the Santa Fe opt-out campaign — enters the legislative record. The concern is now on record from the state's largest district.
“According to their CEO, Amira collected over 600,000 distinct voice recordings from children as young as five across New Mexico last year. Parents were never notified. Parents were never asked for consent. … As the legislators entrusted with protecting our kids, I am asking you all to direct PED to provide all parents with alternatives to the voice recognition.”
A direct answer to day 1, built from day 1's own record — the 600,000 figure and the troubled-child alert are the vendor's and PED's statements, documented above. She also stated the company has raised over $40 million in venture capital from investors including Google and Amazon; her characterizations are her own, cited here as part of the record.
“Simply procuring Amira through CES does not lend itself to LEA participation or full transparency in the decision-making process. … I have personally observed in classrooms across our district widespread challenges with Amira. … Teachers observing that a student's Amira results do not mirror what they observe or other assessment measures, leading to mistrust of the data. … A recent memo from Secretary Padilla requires districts to progress monitor our students using the Amira platform, essentially removing existing district systems already in place. … What is the best way for stakeholders to weigh in on the development of this critical RFP?”
The first district administrator to corroborate, on the legislative record, the field problems committee members had raised the day before — and the first public reference to a Secretary-level memo requiring Amira for progress monitoring. That memo is a discrete public record; the IPRA template on this site now requests it. Roswell, she said, is assessing its youngest students twice.
“I am asking you to reflect on the over two decades that we have been extracting our children's data with commercial digital technology as we head into investing in an even more intrusive, time-intensive platform. … Many [teachers] feel as if their job has been reduced to being a test monitor.”
The longest time horizon offered in either day of the hearing — the current deployment placed at the end of a twenty-year arc of digital assessment.
Two days after the hearing, a local creator’s explainer video about Amira drew 578 comments from 402 accounts in five days — among them 10 self-identified educators and 36 self-identified parents or family members, with 88 comments raising privacy, consent, or recording. The most-liked comment drew 2,003 likes. Comments defending the deployment: 0.
Anonymous social comments are not verified fact, and this campaign does not treat them as such. What is countable is the volume, the speed, and the distribution — and that in a thread this size, the count of comments defending the program is zero.
§07What this record changes
The CES umbrella contract, the annual intergovernmental agreement, the state's retention configuration, and the troubled-child-alert documentation are all now named in the PED request template — each traceable to sworn-in-effect testimony with a timestamp.
File an IPRA →PED's own timeline — public notice this fall, award and transition in 2026-27, in front of students in 2027-28 — is the window in which vetting standards, retention defaults, accommodation access, and outcome-contingent terms can be written into the procurement rather than requested after it.
The proposal →Op-ed July 13. Hearing July 15. Vendor privacy-policy update July 20. The policy diff documents the third event section by section; this page documents the second.
The Amira Policy Diff →The first assessment window opens with the school year. A parent who wants the paper or non-verbal format needs the request in writing before it opens — and the hearing record adds the operative detail that the paper form currently requires an IEP.
The clocks →Corrections: if any quotation or timestamp on this page differs from the archived recording, contact the campaign via the contribute page. Corrections are applied and dated in place. This page documents a public proceeding; it makes no claims about any individual beyond their words in that proceeding.