Who We Mean by Vulnerable Students

This is the definition Higher Learning Lab uses across its research, resources, and the Evidence Tracker. It spans early childhood through higher education.

Definition

A vulnerable student is a learner, from early childhood through higher education, whose developmental stage, disability, language, economic or housing circumstances, care or justice-system involvement, identity, educational pathway, or health and social circumstances make at least one of the following more likely:

Greater susceptibility. The student is less able to judge a tool, program, or adult’s accuracy, purpose, or limits, or more open to forming a relationship with an AI tool.
Greater reliance. The student has fewer human sources of instruction, care, or connection, so any single source of support, including AI, is more likely to substitute for those relationships than to supplement them.
Unequal benefit or harm. Gains from a tool or program are smaller, or harms larger, than for other students, because of access, design, or fit.
Invisibility in the evidence. The student is under-sampled in research, or results are not reported separately for them, so the evidence cannot say whether findings apply to them.

Vulnerability describes circumstances and risk, not deficits in the student. Categories overlap, and many students belong to several.

Why the fourth condition matters

A study that does not say who was included, or does not report results for subgroups, cannot tell educators whether its findings hold for the students they serve. That is why HiLL treats vulnerability partly as a reporting question: our appraisals record whether a study describes its sample on these dimensions and whether it reports results for them, and our gap register tracks where that evidence is missing.

How the definition was built

U.S. statutory definitions that schools and colleges already report against: the ESSA accountability subgroups, the Perkins V “special populations” (which cover secondary and postsecondary learners), IDEA and Section 504, the McKinney-Vento definition of homelessness, Title I Part C (migratory children), and the Higher Education Act’s definition of first-generation students.
Higher-education research categories for students the K–12 categories miss: NCES’s nontraditional-student characteristics, student parents, basic-needs insecurity, and international students.
Evidence on AI and attachment identifying who is most susceptible to relationships with AI: the American Psychological Association’s health advisory on AI and adolescent well-being, Common Sense Media’s research on teens and AI companions, and the OpenAI–MIT Media Lab study of affective ChatGPT use.

The 24 categories

Developmental stage

Category	Who it includes	Level	Basis
Young children	Pre-K through grade 3 (roughly ages 3–8)	K–12	Attachment is still forming at this age
Adolescents and emerging adults	Ages 10–25	Both	APA (2025) defines adolescence as 10–25 and calls for added safeguards from AI

Disability and learning differences

Category	Who it includes	Level	Basis
Students with disabilities receiving special education	Eligible under IDEA, with an IEP	K–12	IDEA, 20 U.S.C. 1401(3); 34 CFR 300.8; ESSA subgroup
Students with disabilities under Section 504 or the ADA	A disability that substantially limits a major life activity; a 504 plan or college accommodations	Both	Section 504, 29 U.S.C. 794; ADA; Perkins V
Neurodivergent students	ADHD, autism, dyslexia, or other learning differences, with or without a formal plan	Both	Sample-composition reporting standard

Language

Category	Who it includes	Level	Basis
English learners and multilingual students	Identified English learners in K–12; in higher education, students studying in English as an additional language	Both	ESSA subgroup; Perkins V

Economic and housing circumstances

Category	Who it includes	Level	Basis
Students from low-income households	ESSA “economically disadvantaged” in K–12; Pell-eligible or low-income in higher education	Both	ESSA subgroup; Perkins V
Food- or housing-insecure students	Food or housing insecurity on validated survey measures	Both	Hope Center #RealCollege survey: 41–48% food insecure and 48–60% housing insecure in fall 2018, depending on college type
Students experiencing homelessness	No fixed, regular, and adequate nighttime residence, including doubling up, motels, shelters, and cars	Both	McKinney-Vento Act, 42 U.S.C. 11434a(2); Perkins V
Migratory students	A qualifying move within the past 36 months as, or with, a migratory agricultural worker or fisher	K–12	ESEA Title I Part C, 20 U.S.C. 6399(3)

Care and system involvement

Category	Who it includes	Level	Basis
Foster youth and foster care alumni	In, or aged out of, the foster care system	Both	Perkins V; ESSA report cards
Justice-involved students	Students in neglected or delinquent facilities or with juvenile justice involvement; incarcerated or formerly incarcerated college students	Both	ESEA Title I Part D; Pell Grant access for prison education programs restored July 1, 2023
Military-connected students	Students with a parent on active duty; student veterans and service members	Both	Perkins V; ESSA report cards

Identity-based marginalization

Category	Who it includes	Level	Basis
Racially and ethnically minoritized students	Students from racial and ethnic groups underrepresented or underserved in their context	Both	ESSA subgroup
LGBTQ+ students	Students who identify as lesbian, gay, bisexual, transgender, queer, or questioning	Both	CDC YRBS 2023: 65% of LGBTQ+ high-school students reported persistent sadness or hopelessness, vs. 31% of cisgender heterosexual peers
Immigrant, refugee, and undocumented students	Immigrants, refugees, asylum seekers, and undocumented or DACA students	Both	Plyler v. Doe (1982) guarantees K–12 access
Rural students	Rural locale as classified by national statistics (in the U.S., NCES locale codes)	Both	Access and connectivity differences

Educational pathway

Category	Who it includes	Level	Basis
First-generation college students	Neither parent completed a bachelor’s degree	Higher ed	Higher Education Act, 20 U.S.C. 1070a-11(h)(3)
International students	Studying outside their country of citizenship on a student visa	Higher ed	Open Doors 2025: 1,177,766 international students in U.S. higher education in 2024/25
Student parents	Students caring for dependent children, especially single parents	Higher ed	GAO-19-522: about 22% of undergraduates were parents in 2015–16, 55% of them single parents
Nontraditional students	Delayed enrollment, part-time study, full-time work, financial independence, dependents, single parenthood, or a GED	Higher ed	NCES nontraditional-undergraduate characteristics
Students with low prior achievement	Below proficiency or in the lowest achievement group, as defined in the study	Both	Sample-composition reporting standard

Health and social connection

Category	Who it includes	Level	Basis
Students with mental health needs	A diagnosed condition or elevated symptoms	Both	CDC YRBS 2023: 53% of female and 65% of LGBTQ+ high-school students reported persistent sadness or hopelessness
Lonely or socially isolated students	High loneliness, weak social support, or insecure attachment	Both	OpenAI–MIT Media Lab (2025): people with a stronger attachment tendency were more likely to report negative effects of chatbot use (adults; not peer-reviewed). Common Sense Media (2025): about a third of teens had chosen an AI companion over a person for a serious conversation

Using this definition

Research describes groups, not individuals. HiLL codes a study’s sample and subgroup results; it never holds data about individual students.
Outside the U.S. The statutory definitions above are U.S.-based. For other countries, we use the closest local equivalent and say so.
The list will change. Revisions are versioned and documented. Suggestions are welcome at [email protected].

Suggested citation

Higher Learning Lab. (2026). Who we mean by vulnerable students. https://www.higherlearninglab.org/resources/vulnerable-students/