Below is a How To Refuse GenAI guide that Illinois Families for Public Schools created for Chicago Public Schools (CPS) families. But we think much of the information here will be relevant and helpful for families outside Chicago as well, including the list of products in use in CPS.

There is more background info on genAI use in CPS here. Unlike in NYC and Los Angeles, there has not been any moratorium placed on genAI products and models in Chicago.

For additional resources on refusing ed tech, we recommend Emily Cherkin‘s work, Screens Beyond Schools, and Fairplay’s Screen Time Action Network. Reach out if you have questions or need advice: info@studentprivacymatters.org. 

Why and how Chicago Public Schools families can protect children from Big Tech’s experimental data grab

[Download this document to print here]

What is genAI? Many digital products, sites, software and apps, including those used for educational purposes, now incorporate algorithms derived from or “trained on” statistical patterns extracted from very large datasets. This technology is commonly referred to as artificial intelligence (AI). Some AI products, known as generative AI (genAI), use statistical patterns as the basis for software programs that can generate text, including computer code, as well as audio, images and video. GenAI is an experimental technology, which has only been commercially available since 2022.

Why should families be wary of genAI use in schools? 

• There is growing evidence that genAI use harms student learning.
• It also poses major risks to the environment, to the mental health of children and adults and to the security of computer systems.
• And genAI companies profit unethically and possibly illegally from the input datasets that their products’ trained on, the bulk of which is made up of the work of creative artists of all kinds. Other genAI training data sources include information generated by product users, and in the case of ed tech, that means data from children and young people in schools.

What do vendors claim these products can do? Software incorporating genAI includes products marketed as able to generate college recommendation letters in seconds, tutor children in K-3 in reading, monitor how often students are using the bathroom, detect “drops in Wellness Level,” surveill classrooms for teacher evaluation, and more.

Do any public schools prohibit genAI use? Both of the country’s largest public school districts, New York City and Los Angeles, recently placed a moratorium on genAI use. Based on an extensive investigation, the Brookings Institute has concluded that “at this point in its trajectory, the risks of utilizing genAI in children’s education overshadow its benefits.” And earlier this year more than 260 organizations and experts issued a statement calling for “a five-year pause on genAI products for students in preK-12 schools.”

Do privacy laws protect students’ data from being used as input for genAI models? No. Unfortunately, state and federal laws have limited protections for student data, with many loopholes allowing tech companies to profit from data they collect when students and school employees use their products. For example, Illinois’ Student Online Personal Protection Act (SOPPA) only protects students’ identifiable data from being exploited by companies training their statistical models for genAI products. However, it is nearly impossible to anonymize personal data, so this is a dangerous exception.

What can families do? First, contact your school board member and ask them for a moratorium like NYC and LA. In the meantime, you can refuse genAI products for your child. Note that Chicago Public Schools’ AI Guidebook supports this, stating “[p]arents and guardians can choose for their child not to use specific GenAI tools“ (p. 19) so that “your child’s educational experience aligns with your family’s values and preferences.“(p. 25) See below for a sample letter to use to notify your school of your refusal.

What products in CPS utilize genAI? We aren’t sure! We have compiled a partial list of products and vendors in use in CPS that may be utilizing genAI (see below.) However, the information CPS provides about the hundreds of tech products approved for use is incomplete, with some products not listed, and others missing contracts or information about what data is being shared. Also, the district’s database does not necessarily tell you whether a product includes genAI, and some companies have embedded genAI functions that aren’t described in their contracts or on their websites.

How do I find out what genAI products my child uses or are used to process my child’s data? Ask your child’s school. Under Illinois state law, you have the right to inspect and correct any of your child’s personal information held by your child’s school or their vendors. So, your school must be able to provide you with a list of all products that school employees use to collect, process or share data about your child or that your child uses directly

What about biometric information? Some AI and genAI products process biometric information. Biometric information is the most sensitive type of personal data because it is irreplaceable if exposed or stolen. It includes things like fingerprints and retina scans. The Illinois School Code (105 ILCS 5/34-18.34) says schools must get prior written consent from parents if students’ biometric information is collected or processed. Note that biometric information may only be used in schools for identification or fraud prevention, not simply for instructional purposes.

Some products in use in CPS, including Amira, Lexia English and Connect to Literacy, record and use students’ speech. Voice recordings can be transformed into voiceprints, which are biometric information according to the IL School Code. And the FTC has said that voice recordings themselves are biometric information.

What other concerns might parents have about tech use in school? GenAI technology is the newest push from Big Tech in our schools. But it is a continuation of issues we were already seeing: overuse of digital devices, inappropriate data collection, surveillance of students and staff, commercial companies shaping curriculum and instruction, and tech used as a substitute for skilled professional educators who build relationships with students as a crucial part of teaching them.

Where can I find out more about these issues? This reading list on AI in schools is a great place to start. Organizations like Fight for the Future, Schools Beyond Screens, the Parent Coalition for Student Privacy, the AI Moratorium Coalition for NYC Schools and Distraction-Free Schools also have many helpful resources. IL-FPS released a Pledge to Protect Human Centered Education in August with a list of references on these issues. More questions? Reach out info@ilfps.org or 773-916-7794

SAMPLE REFUSAL LETTER

---

Products we’ve flagged for genAI

Under SOPPA, vendors must enter a written agreement with public school districts in Illinois before they can collect and process student data. That agreement must be posted on the district website, and it must include certain information, like subcontractors that the vendor is sharing student data with and a list of data elements that the vendor is collecting.  (It is important to note that SOPPA applies to both for-profit and non-profit vendors, and it applies whether or not students use the product directly.)

You can find CPS’ database of district-approved products, which they provide to comply with SOPPA’s requirements here.

This database is not complete (i.e. it violates SOPPA’s requirements!) At a minimum it is missing several sites and products that are widely used in CPS, including Aspen, the district’s student information system; Khan Academy; collegeboard.org; and my.ACT.org. In addition, the majority of vendors are listed without an accompanying data-sharing agreement, and many data-sharing agreements do not provide full information about what data elements are being shared with whom.

(You can also see CPS’ list of pre-qualified ed tech vendors for 2026 here; pre-qualified vendors must then enter into an agreement with the district. Some contracts with vendors can also be found on the CPS Department of Procurement website’s Contracts Awarded page, but as of September 20, 2026, that page is down.)

Adobe Creative Cloud Express	Includes GenAI functions for generating text and images, known as “Adobe Firefly”; these features are turned on by default. No data sharing agreement posted in the CPS database for this or any other Adobe products, nor is there any information about what features are permitted within CPS.
Amira Learning	Uses genAI and speech recognition to process students’ voice data. There is no mention in the CPS data sharing agreement of biometric information collection or artificial intelligence usage. New York City’s public schools included Amira in the genAI product ban. New Mexico Public Education Department now allows school districts to opt out of using Amira for assessment after parents have raised concerns about its use in New York City and New Mexico. According to their website, Amira uses Anthropic’s Claude platform.
Branching Minds	Used districtwide for tracking multi-tiered systems of support (MTSS), a system of monitoring students esp. those with behavioral or academic challenges. Includes data like social-emotional assessment results and information on preK students and students in temporary living situations. Uses genAI. No agreement posted in the CPS database.
Brisk	Shares data with Anthropic, OpenAI and GPTZero AI detector. GenAI interface for students (Brisk Boost) and teachers (Brisk Teaching). Listed on district genAI Pilots website.
Carnegie Learning	MATHia is Carnegie Learning’s “intelligent” tutoring system for math. It uses machine learning (statistical algorithms), but it is not clear from the vendor’s website whether it incorporates an genAI or LLM technology to generate text for written feedback for students at this point. A paper from researchers at the company discuss this as something in the works currently as of July 2025. The data sharing agreement links to a list of subprocessors, but there is no explicit information about known genAI products listed here although AWS does include the ability to create genAI tools. CPS is also using MATHia Mathstream, gamified math videos, from Carnegie Learning.
Class Companion	Interface that uses genAI for grading student work. Not approved CPS software, but possibly being used in CPS schools. Shares data with OpenAI.
DooWii	Shares data with Google for genAI, as well as Anthropic and OpenAI. According to data-sharing agreement, DooWii is a “K-12 conversational analytics platform that unifies data from existing CPS systems (Aspen, Skyline, Google Classroom, ERP , GoCPS, etc) and lets educators, staff, and leadership explore data in natural language for role-based insights to support teaching, learning, and district-wide decisions” and includes an optional “Predictive Modeling Add-On… [for machine learning]-powered forecasts (e.g., risk, outcomes, engagement patterns).”
Edpuzzle	No data sharing agreement on CPS site, but according to vendor website, it shares data with OpenAI “for certain AI-enabled functionalities of the Edpuzzle Teacher Assist service”
EPS	EPS’ Reading Assistant uses AI, including voice recordings according to p. 265 of the data sharing agreement (linked underneath the Wordly Wise entry in the CPS database), which also says that the application shares data with Amira Learning (p. 266). The agreement describes Reading Assistant as “an AI-powered reading tutor that listens to students as they read aloud, providing real-time feedback and personalized tutoring.”
Gemini by Google	Listed on the approved product page, but the data sharing agreement is from 2020, predating the effective date of the major amendment to SOPPA and the existence of Gemini and Notebook as Google products. Also listed as an approved generative AI product in the CPS AI Guidebook, but only for people older than age 18. (Note: This product was previously named “NotebookLM”.) Essentially no information in the posted agreement about what data Google is collecting and processing, nor what additional parties they may be sharing it with. According to an earlier version of the CPS-approved software database, teachers were permitted to input student data into Gemini. According to a district IT Department document, a “broader rollout” of Gemini to all high school students was cancelled at the start of the 2026-2027 school year, but “the district has shifted to a future-readiness model, working directly with individual schools on potential rollouts tailored to their specific needs.”
Gemini Notebook	Listed on the approved product page, but the data sharing agreement is from 2020, predating the effective date of the major amendment to SOPPA and the existence of Gemini and Notebook as Google products. Also listed as an approved generative AI product in the CPS AI Guidebook, but only for people older than age 18. (Note: This product was previously named “NotebookLM”.) Essentially no information in the posted agreement about what data Google is collecting and processing, nor what additional parties they may be sharing it with.
HMH	The HMH data sharing agreement, which applies to more than 40 products, states that HMH shares information at least 23 subcontractors, including Amira Learning (see above), Eloqua (possibly an Oracle product for marketing which utilizes genAI). HMH products Writable and Waggle both use AI according to the HMH website, but no information on what this entails is included in the HMH Data Sharing Agreement. The HMH website says they have an “arrangement” with OpenAI.
Imagine Language and Literacy	The data sharing agreement with Imagine Learning includes nine different products and lists 29 subcontractors, including Microsoft Azure OpenAI. It is unclear what “student-generated work” is being shared with OpenAI (p.33).
Kami	No data sharing documents, but their privacy policy says they are sharing data with Google Gemini. Kami Coach, a new product, appears to be a genAI-based tutoring interface. Other Kami products may be recording students’ voices.
Khanmigo	Khan Academy is in the district’s pre-qualified ed tech vendor list, but is not listed in the approved software list, nor is its AI tool, Khanmigo.
Lexia	Uses AI and, for Lexia English, speech recognition according to the vendor website, but there are no details about this in the data sharing agreement, other than indicating that “student generated content” is collected.
MagicSchool	In the CPS approved list, but no data sharing agreement posted. According to the vendor website, it shares student and employee data with more than two dozen subprocessors, including, for genAI functions, OpenAI, Anthropic, Google Gemini, Adobe, Amazon OpenAI.
Magma Math	Shares students’ classwork with OpenAI and/or Google (likely Gemini) for genAI processing. Based on the vendor website, it appears that the product records and processes a student’s handwritten input on a tablet.
Microsoft CoPilot	Appears in the district’s approved software database but with no data sharing agreement available. It is also listed as an approved generative AI product in the CPS AI Guidebook. According to an earlier version of the CPS-approved software database, teachers were permitted to input student data into CoPilot.
NewsELA	NewsELA has multiple products, but there is no contract on the CPS website to indicate which are in use in the district. The vendor’s privacy policy says that NewsELA is sharing personal information with OpenAI, Google and Anthropic to “develop AI-powered applications.” It also says that “To the extent a student engages with any AI Tools through the Services, this includes disclosing Student Data to our AI Vendors that enable such AI Tools, in accordance with our contractual terms with such AI Vendors and our Terms of Use.”
Otus	Approved, but lists no data sharing agreement and no information about the generative AI functions in the tool. The company has an AI Tool Terms of Use page which instructs users to not enter personally-identifying information into the application, even though the purpose of the software is to process and generate content that contains PII. The Otus website includes a blog post about using one of Otus’ assessments on Anthropic’s AI chatbot/agent Claude.
Parlay	Data sharing agreement is posted, but their privacy policy includes a much longer list of third-parties whom they share with including OpenAI. The Parlay Ideas Genie function is integrated with OpenAI’s ChatGPT product. Parlay is listed in the CPS’ Instructional GenAI Playbook as one of several “Active Generative AI Pilots.”
Playlab	The CPS data sharing agreement with Playlab says that it results in custom AI-tools for classroom and administrative use and mentions that it integrates “multiple AI models”. However, neither the agreement nor any links on the Playlab website provide any information about subprocessors or subcontractors that Playlab is sharing student data with.
Quill	Quill has no data sharing agreement listed in the CPS database, but its entry listing says: “By using Quill, teachers are able to implement research-based expository writing instruction in their classrooms. Quill strives to improve outcomes for every single student who does not have consistent access to the finest writing instruction. By automatically grading the writing as they work, students are able to receive instant feedback and personalized instruction. This tailored experience enables students to learn faster, and it saves teachers hundreds of hours spent on grading. Quill currently provides four writing tools that enable students to build writing, grammar, and proofreading skills.” Note that CPS’s own guidance for instructional staff says, in contradiction to this: “Feedback generated by AI should always be checked and supplemented with educator input to maintain the importance of a human-centered educational experience that is built on the relationship between teacher and student.”
SchooLinks	District’s primary college-and-career software, but no contract is listed in the approved software database, which says that it uses AI. Here is SchooLinks’ own description of how AI, including generative AI, is embedded in its functions for staff and students. SchooLinks provides an LLM for school employees to write recommendation letters: “Generates a first draft from the student profile and brag sheet data. Staff reviews before sending. — Draft in seconds — not 45 minutes — Grounded in verified student record data — Staff edits and approves before send.”
Securly	Securly is software for surveilling students online, filtering school network internet access, hallway pass tracking, school-issued device management and school scheduling. Several Securly applications, including its online surveillance product, Aware, disclose data to OpenAI. In total, Securly applications share data with almost three dozen other subcontractors.
Sibme	Teacher evaluation and coaching software that uses genAI and recordings and also includes an application for genAI lesson plan writing. The vendor website describes processing video and audio recording of classrooms with “optional” biometric information. The data sharing agreementsays only that “student work” is shared with OpenAI, but provides no information about video and audio recordings or biometric information collection.
Snorkl	According to their privacy policy, Snorkl shares student data, including voice recording and typed and handwritten text, with Anthropic and OpenAI.
Summit K12	Summit K12’s Connect to Literacy ELD appliction is for English-language learners to prepare them for federally-required assessments. The product assesses both spoken and written language, and from the vendor website, it is capturing and processing students’ voice data. The data sharing agreement only says student-generated data is being shared with AWS (Amazon Web Services), with no details on whether biometric information is being processed nor any information on training of generativeAI models. Their website says that their “Large Language Model AI [was] trained and calibrated using tens of millions of K-12 MLL student responses collected over the past 6 years.”
Summit Learning	Product that provides online, digital instruction in core classes, which is associated with the Summit Public Schools charter chain. CPS provides no data sharing agreement, and the vendor website’s privacy policy has no information about what data is being collected or shared. It appears that genAI may be incorporated into the Summit Learning platform, and Summit employees have described Summit as an “AI native school model,”
Thinkcerca	Provides generative AI feedback for students on their writing as well as plagiarism detection, but CPS data sharing agreement lists no subprocessors or details about generative AI models. Their website terms of service and privacy policy has no info on third-party subprocessors either.

Statement on the NYS Comptroller’s audit of NYC’s Privacy and Security of Student Data

May 4, 2025

The audit from the State Comptroller’s office released today confirms what many NYC advocates have long known:  the privacy policies and practices of the NYC Dept. of Education are sloppy, irresponsible and show a lack of concern for keeping students’ personal information safe from breach and misuse.    This makes DOE’s insistent push to rapidly expand the use of Artificial Intelligence tools in our schools unwarranted, given how these tools represent an even greater risk to student  privacy and safety.

Even more troubling is the DOE contemptuous response to the auditors’ findings and recommendations to improve their processes, dismissing nearly each one as unfounded.  Altogether, the audit’s findings reinforce the lack of trust felt by many in DOE’s competence and caring when it comes to protecting student privacy.

The audit’s findings put in question the AI guidance’s assurances on DOE’s ability to keep student data safe

In the recent DOE AI guidance, they repeat over and over that student privacy is rigorously protected through a vetting  process  called ERMA (Enterprise Request Management Application).   Yet the findings in this audit show that  DOE’s privacy processes are inherently defective.   The DOE’s lack of responsiveness and willingness to improve their privacy policies provide yet more evidence  that their rush to expand the use of AI in our schools is reckless.   AI products represent a special risk to student privacy as many  data-mine personal data to improve their products, which violates the state student privacy law, Ed Law 2D, the NY State Student Privacy law passed by the legislature in 2014.

The audit’s findings, as well as repeated data breaches of NYC student data and its illegal use for commercial purposes reveal the inadequacy of the  DOE’s privacy vetting process.  As a member of the Chancellor’s AI Working Group, I along with other members proposed additional safeguards.  These included independent privacy impact assessments, data security audits, and tests for algorithmic bias that should be required for any educational product using AI.  DOE rejected all these recommendations.   Additional problems with the recently released AI guidance, including DOE’s refusal to rigorously comply with the state privacy law,  are described in our critique here.

The findings confirm DOE’s failure to properly control and safeguard personal student information

The auditors discovered that DOE maintains  no central records as to which vendors and other third parties have access to student personal information, and that they maintain no written policies covering data classification, risk assessment, or backup and recovery, as required by the NIST data security framework specified by Ed Law 2D.

In their response, DOE officials claim  that this conclusion is false, and that they are “able to determine which SIS or other applications that consume student data are in use by a given school or office.”  Yet just last week, on April 28, 2026, the DOE privacy office confirmed in an email to a parent that “at this time, there is no Central list of every educational technology tool used by each school.”

Moreover, according to Ed Law 2D, it is every parent’s right to know which vendors have access to their children’s data, and to receive a copy of the data held by those vendors  within 45 days of their request. Yet this right is chronically  violated by DOE officials, and when parents do receive data files from their vendors, the files can be empty of information.

There are more than 700  companies and other third parties that have access to personal student data according to the DOE website, though the number of the ed tech programs used is likely greater,  as some vendors provide schools with more than one product.   The number of products collecting and processing student data has steadily increased each year, and is even now even more rapidly growing, as DOE adds  new products with AI functionality to be used in classrooms throughout the city. 

Delays in recognizing and reporting breaches

Because DOE officials do not know which schools use which products, they are unable to ensure that when data breaches occur, they are able to inform affected families within the legally required timeline or identify which data elements may have been exposed.

The auditors reported that there were at least 141 breaches of NYC personal student data  between January 5, 2023 through February 27, 2025, and in 48% of cases, the DOE reported them to  NYSED past the legal deadline of 10 days.  In at least one case, it took over 460 days.  DOE also missed the 60 day deadline to inform parents that their children’s data had been breached in at least 11% of the time. [Note: 60 days is in itself too long; NY law requires breach notification by private businesses  and state agencies within 30 days.]

The Illuminate breach and problems with their privacy agreement

Some privacy vendor agreements are never even posted online in violation of the law – like that of Illuminate, which exposed the data of more than a million NYC current and former students in 2022, and yet whose privacy agreement was posted online only after the breach occurred.  Even then,  the agreement hinted that the data was not always encrypted, contrary to the requirements of the law, which turned out to be the case.

The Illuminate example also shows that  DOE does not independently investigate breaches but instead relies on the unreliable reporting of vendors concerning the number and identity of students affected. After the data of more than 800,000 current and former NYC students was breached by Illuminate between late December 2021 and early January 2022, their families were not notified by DOE until March 25, 2022.

Even worse, in May 2024, more than two years after the breach, a  second round of notifications to families revealed that about  380,000 more students and former students also had their information exposed.  This was also seven months after Illuminate had informed DOE of the additional students involved – far exceeding the 60 day deadline in the law, according to the information on the DOE website, which states that they started looking into this matter only after being told by Illuminate that more students were affected in October 2023.  This put additional students and former students at risk of identity theft and more, and unable to promptly acquire the insurance and credit monitoring offered by the vendor for free.

The PowerSchool breach and problems with their privacy agreement

After the massive nationwide breach of the PowerSchool student information system occurred in late December 2024,  parents throughout the country and elsewhere in the state were informed of the breach in early January 2025.  Yet at that time, DOE told a reporter they were still looking into whether any NYC schools or students were affected.

In fact, DOE refused to confirm which schools were involved even after Daily News reported on their names  on February 6, 2025, from information relayed by the State Education Department.  Only after the Daily News reported on this did parents whose children attended these schools receive emails saying DOE was still looking into this matter.  It was not until April  2025  that DOE confirmed to parents that their children’s data had been breached, long past  the 60-day deadline in the law.

To this day, the DOE has refused to post the names of the NYC schools affected by the PowerSchool breach on the webpage that reports on data security incidents, despite guidance from the NYSED that they should do so promptly, to alert the thousands of former students whose data was also exposed and put at risk of identity theft and worse.

As the former NYSED Chief Privacy Officer Louise de Candia wrote on Feb.3, 2025, “ There is no doubt in my mind that PowerSchool violated Education Law Section 2-d and Part 121 of the regulations which require compliance with NIST CSF as well as reasonable administrative, technical and physical safeguards to protect the security, confidentiality and integrity of PII.”

And yet  DOE continues to allow NYC schools to use as many as 16 other privacy-invasive PowerSchool products, including Naviance, which is employed in many if not most New York high schools for college guidance purposes. This is despite the fact that in 2022, it was reported that Naviance  targeted ads for colleges on its student-facing platform disguised as objective recommendations and was shown to allow colleges to discriminate by race by targeting ads only to white students.

More recently, it was announced that PowerSchool had agreed to settle a class action lawsuit  alleging that the Naviance  platform contained ad tracking technology that transmitted a wide range of student data to Google, Microsoft and a company called Heap, including their names, ID numbers, graduation years,  demographic information, photographs and survey responses, as well as  their private communications with teachers.  This would violate not only state privacy laws but also the federal wiretapping statute.   Even now, the DOE has refused to tell parents or students about the Naviance agreement or  inform them they can apply for a portion of the $17.25 million settlement.

The fact that the Illuminate and PowerSchool breaches exposed the data of many thousands of NYC students who had long graduated or otherwise left the system also shows that the data minimization and deletion by vendors required by Ed Lawa 2D is not enforced by DOE. More background here.

To make things worse, the PowerSchool privacy agreement still posted on the DOE website is clearly non-compliant with the law, as it says that the company will only conform to the privacy requirements in federal and state law or in their contract with DOE when it is “commercially reasonable.”

Other problems highlighted in the audit and the DOE’s official response

The Comptroller’s office also found significant weaknesses in DOE’s technical data security controls that should be corrected, including “issues with system monitoring, unsupported systems, and firewalls.” Understandably, the auditors only communicated the details of these security weaknesses to DOE in a separate confidential report.  In their response, DOE makes no commitment to address these technical problems, but instead says that they would address them separately, within the confidential report.

In its response, DOE  claims to have made “several improvements to its privacy practices and policies,” including updating the Chancellor’s Regulation A-820 to “restrict the use of “directory information.”

In fact, the recent amendment to the Chancellor’s Regulation weakened the protections for student data, by redefining  a wide and essentially unlimited range of personal student information, including but not limited to their names, addresses, telephone numbers, email addresses, photographs, grade level, participation in activities and sports, and more, as directory data that can be shared with third parties, even when they are not providing services to schools.  Only an unreliable parent opt out  process was provided to prevent these disclosures from occurring.

Finally, the auditors also revealed that DOE officials took an inordinate time to respond to their requests; and that documentation requests took over five months to fulfill, while requests for meetings took two months  to schedule.

Leonie Haimson is the co-chair of the Parent Coalition for Student Privacy, a member of the NYSED Data Privacy Advisory Committee, the Chancellor’s Data Privacy Working Group and the Chancellor’s AI Working Group

###

April 25, 2026

This one-page summary is also posted as a pdf here.  The DOE deadline for feedback is May 8, 2026 via their survey at on.nyc.gov/AiFeedbackNYCPS

The Guidance itself is posted here; our more detailed critique is here and embedded below, along with a partial list of AI programs currently used in NYC schools.  Also posted online is an annotated pdf of the AI Guidance with our comments.

On the survey, feel free to borrow any of the below points, provide your own, or write, “I urge you to implement a 2-year moratorium, so rigorous protections can be developed to prevent harm to students, including their privacy, their cognitive development, creativity, mental health and the environment – none of which this guidance sufficiently addresses.”

1: Lack of public input

Despite claims to the contrary, the DOE has not been responsive to the concerns of parents or the community in its determination to rapidly expand the use of AI in the classroom. Nor did this AI guidance document receive significant input from those most affected, either students, teachers or parents. Neither the members of the Data Privacy Working Group nor the AI Working Group appointed by the Chancellor Ramos were allowed to provide comment to the guidance before it was released– despite repeated assurances to the contrary from DOE officials.  And from the DOE’s actions, it is apparent that they intend to continue the rapid expansion of AI whatever the official feedback process consists of in the coming weeks.

2: There is no transparency about which AI products can be used, or that when AI is used at all, there needs to be full disclosure

The DOE AI guidance provides no clarity or transparency about which AI products can be used with students, or those that have gone through the DOE privacy vetting process known as ERMA. While the AI Working Group asked for the names of approved products that are currently used in schools, DOE officials refused, saying they had non-disclosure agreements with their vendors. Perhaps as a result, teachers continue to assign students to use off-the-shelf AI products that data-mine personal student information to improve their products –a commercial use specifically prohibited by the state student privacy law. Regardless of which AI tool is used by teachers or students, there needs to be full disclosure as to which program is being employed and for what purpose.

3: The AI guidance fails to rigorously protect student privacy

The DOE privacy vetting process is ineffective and primarily composed of a series of boxes which vendors are merely asked to check off in order to be approved.  This process has not worked to protect student privacy, as shown by recent breaches of personal information of over one million NYC students and the continued illegal use of student data for commercial purposes, as indicated by recent court settlements and consent decrees. Although AI represents an even higher documented risk to student privacy and safety, the DOE has developed no additional privacy safeguards for its use – despite recommendations from the Chancellor’s appointed AI working group and others to strengthen this process.

4: The AI guidance is inadequate, often confusing and even contradictory as to how teachers and students should use the technology

Instead, if offers a traffic light metaphor, with most potential applications in the “yellow” category, meaning used with caution, leaving it up to teachers to use their best judgement in most of these cases, without giving them clear direction. Other directives are contradictory – as to whether AI can be used for student placement. One bullet point says no; the other says placement decisions can be overridden by teachers or students – but how can that be done if there’s no clarity that the decision was made by AI in the first place? Many of the thorniest questions as to the proper and safe use of AI are punted, to be dealt with at some unspecified time in the future.

5: There is no attempt in the AI guidance to address many of the most serious concerns that parents and educators have about AI use

Growing evidence shows  how AI usage can undermine students’ cognitive development, their acquisition of fundamental skills, weaken their critical thinking and creativity, worsen their mental health challenges, and exacerbate climate change. Yet the guidance does not attempt to address any of those risks. Nor does the guidance provide any answers when it comes to the algorithmic biases often embedded in AI, or the technology’s rampant factual errors, called hallucinations. It also has nothing to say about AI’s tendency towards sycophancy— in which AI chatbots have been designed to agree with the user’s opinions, flatter them, and encourage them in whatever course they are considering, no matter how dangerous it may be. All of these are well-known problems with AI and in the latter case, it has even contributed to teen suicide, according to several ongoing lawsuits. The DOE claims that it will address some of these issues by the end of the year, but there needs to be a moratorium now, so that rigorous protections can be established with public input before the use of AI is further expanded in our schools.

For more information, email us at info@studentprivacymatters.org or check our website at www.studentprivacymatters.org Also, please  sign the AI Moratorium Coalition petition at  https://tinyurl.com/petitionAImoratorium  to be kept up to date on this issue.

“

Update: More on the settlement here.

April 2, 2026

It was recently announced that as part of a class action court settlement, the ed tech company PowerSchool and the Chicago Public Schools agreed to pay a total of $17.25 million to students whose privacy was violated by Naviance, a college advising company acquired  by PowerSchool in 2021. In turn, PowerSchool was bought by Bain Capital for $5.6 billion in 2024.

The lawsuit alleged that the Naviance platform contained ad tracking technology that transmitted a wide range of personal data to Google, Microsoft and a company called Heap, including student names, ID numbers, graduation years,  demographic information, photographs and survey responses, as well as  their private communications with teachers.

These practices, the attorneys argued, amounted to  “unlawful wiretapping” and “eavesdropping,” in violation of several federal and state privacy laws.

Naviance is widely used in schools throughout the country for college application and advising purposes, including in many NYC high schools.  Any student who logged into this platform at least once at school or at home between August 18, 2021 through January 23, 2026 is eligible for payment through the court settlement. A preliminary estimate by the attorney is that each student may receive about  $50, depending on how many apply.

You (or your child if they are over 18) is supposed to have been sent a notice by snail mail or email already on how to file a claim as part of the court settlement, along with a Class Member ID number.  But if you haven’t received this notice, you can still submit a claim here.

We have long been concerned about the privacy and safety of PowerSchool programs in general and Naviance in particular, and we have communicated our concerns with DOE’s Chief Privacy Officer, to no avail.

Several years ago, we had shared reports in the publication The Markup, showing how Naviance had been found to allow colleges to send targeted ads to students through its platform, in some cases ads that discriminated by their race.  These ads were purportedly disguised as objective college recommendations.  Using personal data to send targeted ads violates the provisions of the NY Student Privacy Law.

Then, as you may recall, in December 2024,  a massive breach of the PowerSchool student information system exposed the personal data of millions of students nationwide, including  thousands of current and former NYC students.  As a result of this breach, the company has been sued by  many states and districts for failing to implement the most basic data security and privacy protections.   After this occurred, I again urged DOE to cancel its contracts with PowerSchool, which offers many different, highly invasive programs to NYC schools, but received no response.

If your child uses Naviance, beware of any recommendations or other communications that they may receive through this platform.

I’d appreciate it if any parents whose child currently uses the platform might help us investigate the way Naviance works in more detail, to assess whether the company may still be continuing to violate our privacy laws and basic ethical standards, including through their new AI-powered chatbot called “PowerBuddy”.  If you and your child are willing, please email us at info@studentprivacymatters.org..  Please also let us know if you or your child has not received notice of this settlement, so we can inform the plaintiff’s attorneys.

Finally, whether or not you receive a settlement payout, it would be great if you would consider donating to Class Size Matters, earmarked to help fund the Parent Coalition for Student Privacy. Our amazing PCSP co-chair, Cassie Creswell, executive director of Illinois Families for Public Schools, worked with the attorneys on the class action lawsuit and helped identify the original plaintiff. We could really use your support.

This presentation could be used in any school, district or state, but it was created for NYC schools.