August 20, 2025

Zachary Rogers
U.S. Department of Education
400 Maryland Ave SW, Room 7W213
Washington, DC 20202-6450

Re: Comment on Proposed Priority and Definitions-Secretary’s Supplemental Priority and Definitions on Advancing Artificial Intelligence in Education Docket ID ED-2025-OS-0118

Dear Mr. Rogers:

Thank you for the opportunity to provide comments on the U.S. Department of Education’s supplemental priority and definitions on advancing artificial intelligence (AI) in education. This topic is of great importance and timeliness given the growing utilization of generative AI technologies by students and teachers across the country. However, the current priority puts too great an emphasis on the rapid implementation of AI technologies throughout educational settings and too small an emphasis on the necessary legal, ethical, and equity considerations necessary to ensure AI is responsibly implemented throughout school systems. The comments below outline both opportunities for federal investment and suggested revisions to the proposed priority to ensure education stakeholders are adequately equipped to procure, implement, and evaluate AI technologies in their local educational contexts.

Supporting Foundational Research and Evidence Should Be a Top Priority

The proposed priority envisions widespread integration of AI tools into classroom settings to personalize instruction, give students access to advanced tutoring or virtual assistant teachers, and bolster services for children with disabilities, among other applications, usually with the aim of improving student outcomes and experiences. While some of these applications may exist in limited capacities today, and could certainly exist in the future, robust evidence of AI tools’ effectiveness and safety remains limited. There are currently no compendiums of vetted, high-quality AI-driven education tools, no established best practices or promising practices for effective AI implementation, and minimal research linking AI usage to student outcomes. Overlooking these critical research and development needs in the finalized priority — or loosely capturing them in a single proposal to “build evidence of appropriate methods of integrating AI into education” — presents too great a risk for our nation’s students. Priorities should encourage, or require, grant recipients to evaluate the impact of AI implementations in classrooms to begin building this research base. Additionally, whether in these priorities or in future guidance, the Department should call for more federal investments around research and development at the intersection of AI and education.

AI Literacy Must Begin with Educators

The Department is right to focus so heavily on the need for AI literacy, especially for students. With that said, we hope to see more attention placed on educators in the final priority. Teachers and school leaders need clear, accessible training to develop the knowledge, skills, and competencies needed to safely use and understand AI and critically evaluate the implications, limitations, and ethical considerations of AI use. Educators must receive AI literacy trainings that center around informed data privacy practices, and the subsequent legal risks of data misuse. The trainings should also offer strategies to combat the systemic and often unfair discrimination of large language model (LLM) outputs against people of color and women caused by biased training data, especially when utilizing AI for student-facing tasks, and methods to effectively use AI as a part of instruction.

Increase Clarity On How Computer Science Support Will Be Targeted

Expanding access to computer science education across K-12 is an important goal, one that should continue being emphasized, but the priority would benefit from increased specificity on how the Department will target its support. Equitable access to computer science courses and high-quality STEM educators remains limited for rural and urban communities, districts that serve high concentrations of students from low-income backgrounds, and small schools with fewer than 500 students, and female students are less likely to enroll in and complete available computer science offerings. Targeting support towards the districts and populations most in need would offer the greatest impact for mitigating disparities in access.

Add Robust Proposals Around Data Privacy and Ethical Guardrails

Notably absent from the proposed priority is any meaningful exploration of data privacy or ethical guardrails. AI use in classroom settings presents real data privacy concerns, as tools must comply with existing data privacy laws like FERPA and COPPA. Teachers and school leaders grappling with decisions around AI use may not understand these risks, due to unfamiliarity with generative AI technologies and how they store data, leading to misuse and potential violations of federal or state laws. It should be a priority of the Department to ensure schools and districts have the resources to understand relevant data privacy stipulations of AI technologies before mass adoption across their systems.

Additionally, there are a slew of ethical issues surrounding AI use, ranging from its significant carbon footprint to the unsolicited use of copyrighted and private information to train generative AI models. Every school, district, and state education leader should at the very least know about these issues, so they can make informed decisions around AI procurement, adoption, and implementation in their own local context.

Priorities Must Address Improper Use of AI

There are considerable equity considerations surrounding AI use in education, such as systemic bias of generative AI model outputs, that education stakeholders must know about before making policy and procurement decisions. These issues are so prevalent that the Office of Civil Rights authored preliminary guidance to ensure teachers and school leaders use AI in a nondiscriminatory manner. Schools and districts are sure to face legal challenges surrounding the improper and discriminatory use of AI to make decisions that impact students. The Department can play a key role in disseminating resources and guidance to help bolster equitable and responsible AI use, but the current proposed priority shows no acknowledgement that AI technologies present such harms.

Deep Resource Inequities Will Stymie AI Implementation and Adoption

Lastly, deep resource disparities across America’s K-12 education system, especially in schools that serve a high proportion of students from low-income backgrounds and students of color, mean many of the most disadvantaged students will be in schools and districts that cannot allocate any budget toward AI literacy or implementation initiatives. Furthermore, many students lack access to resources that responsible AI implementation will rely on, such as affordable high-speed internet at home, devices (laptops, tablets), or high-quality, representative, and AI literate educators. These are not new issues, but they must be addressed if the Department truly intends to promote widespread integration of AI into education.

Recommendations

For these reasons, EdTrust respectfully recommends that the Department augment existing priorities or create new priorities that:

1. Signal financial support for the research and development ecosystem as they develop foundational research on the impacts of AI technologies on educational outcomes, necessary for identifying promising practices. This can be achieved by:
   1. Creating an additional priority around contributing to the AI research base;
   2. Encouraging, in the priority or separately in grant applications, that recipients dedicate a portion of funding for evaluation purposes; and/or
   3. Including specific reporting requirements to support federal research and evaluation of AI and education related grants that would be used by the Department to report on AI use in schools.
2. Call for funds to support engagement opportunities that connect experts, industry partners, and education stakeholders (including but not limited to teachers, students, administrators, and community organizations) to support definitional alignment and the co-creation of AI procurement, implementation, and literacy resources that best fit the real needs of our students.
3. Support the creation of accessible teacher and leader AI literacy trainings that develop the knowledge, skills, and competencies needed to safely use and understand AI and critically evaluate the implications, limitations, and ethical considerations of AI use.
4. Ensure resources that increase computer science access are targeted at the schools and districts with the greatest need, including but not limited to rural and urban communities, districts that serve a high concentration of students from low-income backgrounds, and small schools with fewer than 500 students.
5. Develop priorities for embedding information on robust data privacy safeguards and ethical and equity considerations into AI resources. This is important to ensure that educators and administrators utilize AI tools in ways that comply with existing data privacy laws and understand the real and potential harms of widespread AI use, like the perpetuation of stereotypes due to biased training data.
6. Acknowledge that, as a necessary component of this broad effort to support AI implementation, the administration must address deep resource disparities facing students of color and students from low-income backgrounds. This may be accomplished by targeting AI in education grants to Title I schools and/or schools and districts with high concentrations of students from low-income backgrounds.

Signed,

American Atheists
EDGE Partners
EdTrust
InnovateEDU
League of United Latin American Citizens (LULAC)
National Urban League
Students Engaged in Advancing Texas (SEAT)