Not sure where to start? To understand the landscape, begin with Know Where and How AI Shows Up in Your Student’s Life. To do something today, jump to the toolkit.
Purpose
Your student is already living in a world shaped by artificial intelligence, whether or not either of you has chosen it. AI increasingly curates their feeds, is used by their teachers, and can be used to help them do homework. This guide is here to help you make sense of how AI has entered students’ lives, and to help your students use AI in ways that are healthy, honest, and age-appropriate.
This guide was designed to be read alongside the Student Guide to AI, which speaks directly to young people about the habits and mindsets behind good AI use. This companion guide draws on research from developmental psychology and the learning sciences to give you two things: a clear picture of where AI shows up in your students’ lives, and practical ways to talk with them about it. Along the way, you’ll find conversation starters you can use at the dinner table, in the car, or wherever the real conversations tend to happen.
All you need to use this guide is curiosity and a willingness to learn alongside your student.
Know Where and How AI Shows Up in Your Students’ Lives
What Is AI?
Artificial intelligence refers to computer systems that perform tasks we typically associate with human thinking: recognizing patterns, making predictions, understanding language, and creating new content. Instead of following a fixed set of instructions written by a programmer, most modern AI learns from large amounts of data and improves as it goes.
Your student most often meets a specific type of AI: generative AI. Generative AI powers tools like ChatGPT, Google Gemini, Claude, and Microsoft Copilot that produce text, images, audio, or video in response to a prompt. Generative AI also powers wrapper tools that provide user interfaces for specific purposes, including teaching and learning. Many of these rely on large language models (LLMs), which are trained on enormous collections of text to predict the next word. That prediction is powerful enough to draft an essay or explain a scientific concept, and it’s also the reason these tools can sound completely confident even when they're wrong and mimic human behavior (points we return to later in this guide).
Key Terms
Generative AI: AI that creates something new, such as a paragraph, image, or song, rather than simply retrieving an existing answer.
Large language model (LLM): the technology behind most chatbots; it generates text by predicting likely word patterns.
Hallucination: a confident but false or made-up answer, including invented facts, fake quotes, and citations for studies that don’t exist.
Training data: the material an AI learned from, which shapes both what it knows and the biases it carries.
This guide uses the same core vocabulary as the Student Guide, and both draw on widely used AI-literacy frameworks from UNESCO (2021), the OECD and European Commission (2025), and the AI Education Project (aiEDU, 2024), so your family shares a common language for these conversations.
¹ Another term for hallucinations is fabrications; this term is used to ensure that AI is not being described in anthropomorphic ways.
Students Use AI Everywhere
Given the heated discourse about AI, it helps to go beyond the headlines and review research. In a nationally representative survey conducted in fall 2025, the Pew Research Center found that 54% of U.S. teens use AI chatbots for help with schoolwork and 57% use them to search for information (Pew Research Center, 2026).
About one in ten teens say they do all or most of their schoolwork with a chatbot’s help. Just as revealing, 59% of teens said that cheating with AI has become a regular part of student life (Pew Research Center, 2026).
AI use has climbed quickly. The share of teens using ChatGPT for schoolwork doubled in a single year, from 13% in 2023 to 26% in 2024 to 54% in 2025 (Pew Research Center, 2025, 2026). Given this trajectory, it is safe to assume that more students use AI today than what the published, out-of-date research demonstrates.
And it reaches well beyond homework. Seven in ten teens have used at least one generative AI tool (Common Sense Media, 2024), and a 2025 study found that nearly 75% of teens have used an AI companion, a chatbot designed for conversation and connection rather than schoolwork (Common Sense Media, 2025).
Here’s the finding that illustrates the need for this guide: many adults don’t realize how much young people in their lives use these tools. In Common Sense Media’s research, about a quarter of parents did not think their child was using AI, even though those same children reported that they were (Common Sense Media, 2024). Closing this adult-belief student-reality gap is the whole point of this guide. You can’t guide what you can’t see, so knowing where AI shows up is the first step to helping your student use it well.
What Makes an AI Tool Safe for Students?
Not every AI tool is built with young people in mind. Some are designed for education, with strong protections in place. Others are general-purpose products that treat a 12-year-old’s data the same way they treat an adult’s (e.g., an ideal customer). Before your student leans on AI, it’s worth knowing what separates a student-safe tool from a risky one. There are five qualities that must be critically evaluated for identifying student-safe AI tools: (1) how it protects data, (2) what safety features it has, (3) whether it supports real learning, (4) whether it provides trustworthy information, and, (5) whether it is transparent about how it works.
It Protects Students' Data
Every time your student types a prompt, uploads a photo, or signs into an AI tool, they hand over information. The most sensitive kind is personally identifiable information (PII): any detail that could be used to identify them. That includes obvious things like their full name, birth date, and home address, as well as their school, location, and photo, and especially sensitive material such as medical or mental health information. With AI tools, there’s an added wrinkle: whatever your student types can be stored, and depending on the tool, used to train the company’s future commercial models, sometimes retained indefinitely. Testifying before the U.S. Senate in 2026, education-technology leader Erin Mote drew a sharp line between tools built for schools, which operate under formal contracts that require compliance with FERPA and COPPA, and general-purpose consumer apps, whose broad public terms can allow student data to be scraped to train future models (Mote, 2026). Nevertheless, not all tools designed for educational purposes necessarily protect student data, so it is important to know what to look for.
Several laws set the floor for how student data must be handled: the Children’s Online Privacy Protection Act (COPPA) covers children under 13, the Family Educational Rights and Privacy Act (FERPA) protects student education records, and Michigan’s Student Online Personal Protection Act limits what ed-tech vendors may do with student data. Appendix E summarizes each in plain language and what it means for you (Federal Trade Commission, 2025; Student Online Personal Protection Act, 2016; U.S. Department of Education, n.d.).
Questions to ask before your student uses an AI tool
Does it use your student’s data for training? Look for a clear statement that inputs aren’t used to train the company’s models or that there is a setting to turn that off (often labeled “don’t train on my data,” “temporary chat,” or “incognito mode”).
Is it built for students? Tools designed for education usually ship with stronger default protections than general-purpose consumer apps (but it is still important to verify this is the case).
Can you actually read the privacy policy? It’s a good sign if there is a plain-language policy that says what data are collected, why, how long they are kept, and whether they are shared. A vague or missing policy is a red flag.
Does it sell data or run ads? Steer away from tools that make money by advertising to your student or reselling their information.
What do independent reviewers say? Independent, family-facing reviewers rate popular apps and AI tools for privacy and safety, including Common Sense Media, the nonprofit Internet Safety Labs, and Mozilla’s Privacy Not Included buyer’s guide (Common Sense Media, n.d.; Internet Safety Labs, n.d.; Mozilla Foundation, n.d.).
At school, just ask. You’re entitled to ask which AI tools your student uses, what data they collect, how their data are protected, and whether you can opt your student out of using the tool.
It Has Meaningful Safety Features
A student-safe tool is built with guardrails. When you’re weighing a tool, there are a few features that are worth looking for:
Content filters. It should refuse to produce sexual, violent, hateful, or self-harm content and decline inappropriate requests rather than play along.
A safe response in a crisis. If a student expresses distress, a well-designed tool responds with care and points them toward real human help rather than engaging with harmful ideas.
Age-appropriate design and age checks that work. Look for tools intended for your student’s age, and be skeptical of a simple “I’m over 13” checkbox. Meaningful safety requires stronger age verification than self-reported age, which young people can easily circumvent. Some places now push further: Australia’s under-16 social media restrictions, in force since December 2025, expect platforms to use layered age-assurance methods rather than self-declaration alone (eSafety Commissioner, 2025).
Honesty about being AI. A safe tool is clear that it’s a machine. It doesn’t claim to be a real person or to have feelings, a tactic some companion apps use to deepen attachment.
Controls and a way to report problems. Well-designed tools for students should have parental and/or teacher controls, adjustable settings, and an easy way to flag bad responses.
It is important to keep in mind that safety features, while beneficial, are not a guarantee. Guardrails can be inconsistent, and harmful material sometimes slips. In one 2025 assessment, Common Sense Media found that a widely used AI assistant’s safety systems repeatedly failed to recognize signs of a teen in crisis (Common Sense Media, 2025). Given these demonstrated inconsistencies, it is important to learn how your student is using AI in and out of school.
Risk assessment resource
Common Sense Media’s Youth AI Safety Institute publishes plain-language risk assessments of popular AI tools, so you can check how a specific product rates on safety and privacy before your student uses it.
It Supports Productive Struggle Instead of Doing the Thinking for Your Student
Learning happens through effort. When a student wrestles with a hard problem, tries an approach that fails, and figures out why, they build understanding that lasts. Researchers call this productive struggle, and it is the mechanism by which learning happens. The risk with using AI is that it can remove the struggle altogether. Ask a chatbot, get an answer, move on. The task gets finished, but the thinking that would have built the skill never takes place. That shortcut has a name: cognitive offloading, or handing mental work to an outside tool.
Some offloading is normal and healthy. You offload phone numbers to your contacts list without losing anything that matters. The concern is what happens when a developing mind offloads the very thinking that school exists to build. Cognitive science research has shown a trade-off for years: offloading a task tends to boost performance in the moment while weakening memory for the material afterward (Grinschgl et al., 2021).
Recent studies highlight how AI use is associated with cognitive offloading. One study showed that heavier AI-tool use was associated with weaker critical-thinking scores, with cognitive offloading acting as the link between the two, and the strongest effects among the youngest participants (Gerlich, 2025). In a widely discussed MIT Media Lab experiment, students wrote essays using ChatGPT, a search engine, or no tools while researchers recorded their brain activity. The AI group showed the weakest neural engagement, reported the least ownership of their work, and struggled to quote essays they had just written, a pattern the researchers labeled “cognitive debt” (Kosmyna et al., 2025). While that study had a small sample size, it provides support for the claim that cognitive offloading is consistent enough to take seriously.
For guardians, a key takeaway is that a tool’s design matters. In a controlled study with high school math students, a general-purpose chatbot that supplied answers actually harmed learning, while an AI built to tutor (one that guided students through problems instead of handing over solutions) improved both grades and retention of the information (Bastani et al., 2025). While both tools used the same underlying technology (large language models and generative AI), students who used them had opposite outcomes. A tool that makes the student do the thinking helps; a tool that does the thinking for them hurts.
The good news is that many students are not simply outsourcing everything. In Common Sense Media’s 2026 survey, 63% of teens who use AI for schoolwork said they get an answer from it in some form, but 77% also said they use it in ways that keep them in the driver’s seat, such as brainstorming a start, checking their work after finishing, or getting feedback on an answer they wrote themselves first. Nearly two-thirds said that when they get stuck, they still try to solve the problem themselves or ask a person before turning to AI. At the same time, 38% said having AI available leads them to come up with fewer of their own ideas, and 39% said they feel like they’re missing out on learning when they let AI complete their assignments (Common Sense Media, 2026). Your student is likely navigating this tension on their own, and you can help them through it.
Signs that an AI tool preserves the productive struggle
Ask questions back instead of just producing the final answer, nudging your student to reason through the next step.
Offers hints and feedback on work the student has already attempted, rather than replacing the attempt.
Adjusts to your student’s level, so the challenge stays productive instead of overwhelming or trivial.
Encourages a “try it yourself first” habit, which the Student Guide describes as making the first draft theirs and not the machine’s.
The goal is not to completely avoid using AI; it is to keep your student in charge of their own thinking, using AI as a spotter at the gym rather than a robot doing the push-ups for them (an image the Student Guide uses, too). When AI supports the struggle rather than erasing it, it can genuinely help personalize instruction to your students’ needs and improve learning outcomes.
It’s Built for Accuracy and Fairness
There are two weaknesses in AI tools relevant to schoolwork: they can be confidently wrong and quietly biased. The Student Guide already teaches your student how to catch both by building habits like verifying facts, asking for sources, and keeping the “human in the loop.” As a parent or guardian, your job is to understand these issues, know how a well-built tool addresses them, and help your student develop an AI checking habit.
The first problem is hallucination: because a chatbot generates text by predicting likely words, it will sometimes produce fluent, authoritative-sounding answers that are simply false, including invented statistics, fake quotes, and citations for studies that don’t exist. The second problem is bias. AI learns from enormous amounts of human-created text, so it absorbs the perspectives from prior text. For example, a 2024 UNESCO study of several widely used large language models found clear, consistent gender bias: the models associated women with words like “home,” “family,” and “children” while linking men to “career,” “executive,” and “business,” and they assigned narrower, more stereotyped roles to people from some ethnic groups (UNESCO, 2024). Left unchecked, tools like these can quietly reinforce stereotypes in the content your student reads and produces.
What student-safe tools do to reduce hallucination and bias
Grounding answers in vetted sources rather than free-associating. Tools built for education increasingly connect the AI to trusted, curriculum-aligned material so answers stay anchored to real content, cites verifiable sources, admits uncertainty, and points the student back to trusted humans.
Keeping a human in the loop. In well-designed classroom tools, a teacher reviews AI-generated content and student feedback before they reach students.
Watching for bias. Responsible developers test their tools’ outputs across different groups of students and fix disparities when they appear.
No tool is entirely free of hallucinations or bias, which is exactly why it is critical to keep human checks a central part of the AI-use process. The most useful thing you can do is echo what the Student Guide asks of your student: treat AI’s output as a first draft to question and not a final answer to trust. When your student brings home something an AI produced, “How did you check this?” is a better question than “Did you use AI?”
It’s Transparent About How It Works
A student-safe AI tool is honest about what it is and what it can and can’t do. You shouldn’t have to reverse-engineer how it handles your students’ data or guess whether they’re talking to a person or a program. Transparency is a common thread running through every major safety framework (for example, UNESCO’s Recommendation on the Ethics of AI, the OECD and European Commission’s AI literacy framework, the U.S. National Institute of Standards and Technology’s AI Risk Management Framework, and the EDSAFE AI Alliance’s SAFE Framework; NIST, 2023; OECD & European Commission, 2025; UNESCO, 2021), and it’s what makes the other four characteristics checkable in the first place.
What transparency looks like in a student-safe AI tool
Disclose that it’s AI to users, so your student always knows they’re interacting with a computer program, not a person.
Explains data practices in plain language, including what it collects, how it’s used and for how long, and who it’s shared with.
Admits its limits, telling users that its answers can be wrong and describing how it catches errors (human review, citations, or pointing students to a teacher).
Says which features use AI, so it’s clear when a lesson, a grade, a piece of feedback, or a recommendation came from a machine.
If a company doesn’t have this information readily available, treat that opacity as a flag for your student to exercise caution.
Use the checklist in Appendix A: Questions to Ask About an AI Tool to size up any tool your student uses, whether they downloaded it or the school assigned it.
Where and How AI Shows Up at School
AI Is Already Built Into the Tools Your Student Uses
When people picture AI in school, they usually imagine a student typing a prompt into ChatGPT. While that happens, a lot of the AI in your student’s school day is less visible because it is embedded in educational technology (edtech). AI now runs inside learning platforms, reading and writing assistants, research tools, tutoring programs, and the analytics dashboards teachers use to track progress. Your student may be using AI at school without either of you thinking of it as “using AI.”
The sheer volume of edtech tools used is an important part of the story. In its 2025 EdTech Top 40 report, LearnPlatform by Instructure found that K-12 districts accessed an average of about 2,980 distinct edtech tools over the 2024–25 school year, with a typical student using around 48 of them and educators about 50. Among teens who use AI for schoolwork, Common Sense Media found that about 7 in 10 reach for multipurpose tools like ChatGPT, Gemini, Claude, or Grok, roughly 4 in 10 use a reading or writing assistant such as Grammarly, about 3 in 10 use a research tool, and fewer use purpose-built tutors. In other words, “AI at school” takes many different forms and runs along a spectrum, from vetted, purpose-built tools that a district deliberately chooses to general-purpose apps that students access on their own. The characteristics in the prior section help distinguish the two ends of the purpose-built-to-consumer-grade spectrum.
Schools have their work cut out for them in managing AI use because of how this technology arrived. Education technology leader Erin Mote describes generative AI as an “arrival technology” rather than an “adoption technology”: rather than moving through the usual path of pilots, review, and district purchasing, it entered classrooms through students’ and teachers’ personal use before schools could evaluate it (Mote, 2026). Students’ AI use reflects the nature of AI as an arrival technology: 44% of teens who use AI for schoolwork had a tool blocked on a school network or device, and when that happened, 59% simply switched to a personal device to reach it (Common Sense Media, 2026). For you, as a parent, the takeaway is that AI is present in your student’s education, whether or not their school has fully caught up, making your awareness and guidance a critical part of their safety net.
The Benefits: How AI Can Help Learning
When it is used well, AI can have real benefits for student learning. AI helps most when it supports a human teacher rather than replacing one, and when the tool was built for learning rather than borrowed from the consumer market. The evidence base for AI tools is still developing, so treat the benefits below as promising trends and not proven facts.
Here are some of the ways AI supports learning:
AI provides data-based insights for teachers. AI tools help find patterns in large amounts of information, making them useful for turning the messy data of a classroom into something a teacher can act on. For example, tools can flag which students are struggling with which concepts, surface common misconceptions, and give teachers a clearer picture of where a class stands. According to a Gallup-Walton survey, 61% of teachers who use AI said it provides them with better insights into student learning and achievement data (Walton Family Foundation & Gallup, 2025). Reviews of the research have found that AI, which delivers automated feedback to human tutors, can improve both instructional quality and student learning (Fesler et al., 2026). Importantly, these uses keep the human in the driver’s seat: AI doesn’t judge your student; it helps a teacher see your student more clearly.
Smart tutors. A well-designed AI tutor can offer something a busy classroom can’t always provide: patient, one-on-one, immediate help. In a controlled study of high school math students, an AI tutoring tool (guiding students through problems with hints and steps) improved grades and retention (Bastani et al., 2025). A randomized trial of an AI tutor in UK classrooms likewise found that it could support learning safely and effectively when designed with pedagogical guardrails (LearnLM Team, 2025). These findings demonstrate the importance of using AI tools purpose-built for education: a tutor built to teach is very different from a chatbot that answers.
Accessibility and learning-access supports. Some of AI’s clearest benefits are for students who face barriers to learning. AI-powered text-to-speech, real-time translation, and reading and writing support can make material that would otherwise be out of reach accessible. Nearly 60% of teachers in the Gallup-Walton survey said AI improves the accessibility of learning materials for students with disabilities (Walton Family Foundation & Gallup, 2025). Students are already using AI for accessibility: in Common Sense Media’s 2026 survey, teens with an IEP or 504 plan were more likely than their peers to customize AI tools for their specific learning needs, 44% versus 34% (Common Sense Media, 2026). For multilingual learners and students with dyslexia, ADHD, or other learning differences, these tools can be beneficial.
Time savings for teachers. The most immediate benefit of AI falls outside of your student’s direct edtech use. AI can be used to give teachers more time, and in turn, allow them to focus on the most important things for student learning. Teachers spend hours drafting worksheets, adapting materials, and doing administrative paperwork, and AI can speed up much of that work. The Gallup-Walton survey found that teachers who use AI at least weekly save an average of 5.9 hours a week, about six weeks over a school year, time many said they reinvested in individualized lessons, feedback, and communicating with families (Walton Family Foundation & Gallup, 2025). When AI absorbs the routine work, teachers can spend more of their attention on the parts of teaching that only a human can do.
These benefits are real, but they are not automatic, and even enthusiastic teachers are clear-eyed about the trade-offs. In that same Gallup-Walton survey, a majority of teachers worried that frequent AI use would weaken students’ independent thinking (57%) and critical thinking (52%) (Walton Family Foundation & Gallup, 2025). To support your student in developing safe and healthy AI habits, it is important to weigh the benefits of AI use alongside the risks.
The Risks: Where AI Can Undermine Learning
While AI can help students learn, it also carries real risks. This section provides an overview of the risks that most often surface in research on AI use in education. These risks are not ranked or exhaustive, and we deliberately address several major risks elsewhere: data privacy, AI relationships and mental health, and deepfakes are addressed later; hallucinations, bias, and cognitive offloading were addressed earlier. Additionally, many of the risks below are greater when a student relies on a consumer-grade tool rather than one built for learning.
Here are some of the ways AI may undermine learning:
Sycophancy: the AI that tells your student what they want to hear. Most chatbots are trained to please. They learn from human ratings, and people tend to rate agreeable answers more highly than challenging ones, so the models drift toward flattery and agreement. As one expert put it in Senate testimony, consumer models “do not teach; they indulge” (Mote, 2026). A 2026 study in the journal Science tested 11 leading AI systems and found they affirmed users’ actions far more readily than humans did, and, tellingly, that people trusted and preferred the models that flattered them (Cheng et al., 2026). Researchers at Penn State and MIT found that the tendency grows over the course of a long conversation, especially once the tool has stored details about the user, making it more likely to mirror that person’s views (Emmons, 2026). And a growing body of work shows why this matters: a sycophantic AI can raise a student’s confidence in a wrong answer without bringing them any closer to being right (Cheng et al., 2026).
Over time, this can lead to students’ learning misconceptions, which is why the Student Guide urges them to push back on AI and treat its agreement with suspicion. The version for home: when your student says “the AI agreed with me,” treat that as a cue to ask how they checked, not a reason to relax.
Academic integrity. Separate from whether AI builds or erodes skills is a plainer question: is the work honest, and does it still mean anything? AI makes it easy to hand in work a student didn’t do, and the norms around this are genuinely unsettled. In a 2026 Pew survey, 59% of teens said cheating with AI has become a regular part of student life (Pew Research Center, 2026), and Common Sense Media found that most teens who use AI for schoolwork get an answer from it in some form, with a quarter using it as-is (Common Sense Media, 2026). But the line between “help” and “cheating” is blurry, and students often can’t find it: Common Sense reported that about a third of teens don’t understand their school’s AI rules, and that those rules vary from teacher to teacher, making it difficult for them to navigate (Common Sense Media, 2026). Cheating itself isn’t new: one study comparing high schoolers before and after ChatGPT’s release found similar overall rates of cheating, roughly 60–70%, suggesting AI has changed how students cut corners more than how often they do so (Lee et al., 2024). Your student is navigating genuinely murky territory, which makes conversations at home and clear guidance from the school more useful than suspicion or worst-case assumptions.
Self-efficacy, confidence, and curiosity. Self-efficacy is a student’s belief in their ability to complete what is required to be successful (Bandura, 1997). Decades of research trace the development of self-efficacy to mastery experiences: the confidence that grows from tackling something hard and succeeding. Here’s the trap: if AI handles the hard parts, a student misses out on the very experiences that build self-efficacy and confidence. Some researchers describe a risk of “learned helplessness” (Zhang & Xu, 2024), in which confidence is propped up by technology rather than rooted in the student’s own competence. When paired with sycophancy, the result can be a student who feels capable without developing new skills and knowledge, so their confidence evaporates the moment the tool is gone. A related concern is curiosity: some educators and researchers have suggested that collapsing the distance between a question and an instant answer can dull the urge to wonder and explore, which is the very state in which learning takes hold. Students sense these tradeoffs themselves. Researchers at the Brookings Institution found that young people’s leading worry about AI isn’t job loss but “cognitive loss,” the fear of losing their own ability to think (Burns et al., 2026).
2 Much of the self-efficacy research so far focuses on older and university-age students, and the developmental stakes may be even higher for younger children still building foundational skills.
Weaker human connection. When asked which skills will matter most for their future, the teens in Common Sense Media’s 2026 survey put “working well with other people” at the top (Common Sense Media, 2026). Yet an always-available, endlessly agreeable AI doesn’t build critical relational skills the way working with peers does. AI tools can’t effectively teach a student to read social cues, sit with disagreement, or repair a conflict. The more a student turns to a frictionless AI for help, company, or feedback, the fewer reps they get at the harder, messier, human work that develops those abilities.
None of these risks is a reason to ban AI outright, but they do provide reasons to be deliberate about how and when it should be used. Nearly all risks ease when the tool is purpose-built, a human stays in the loop, and the student keeps doing the thinking.
Why Students Reach for AI to Skip the Work
While it’s tempting to view a student offloading their thinking to AI as laziness or dishonesty, it’s usually neither. When a student routinely hands work to AI in a way that skips the learning, it’s most often a signal that something about the work, or their relationship to it, isn’t motivating them to do it themselves. Researchers have studied academic motivation for decades, and their findings point to a handful of recurring root
causes for why a student may not be motivated to do the work themselves. Learning how to support your students’ motivation will help you address the underlying reasons they use AI in maladaptive ways, because telling them to “just do it yourself” rarely works if the underlying reason they turned to AI instead of engaging in productive struggle remains.
“I don’t think I can do this.” A student’s belief in their own ability to succeed is one of the strongest predictors of whether they’ll attempt hard work or avoid it. A student who doubts they can do the assignment has little reason to struggle through it, and AI offers a graceful exit. The productive response isn’t reassurance that they’re “smart”; it’s helping them string together small, real successes that rebuild the sense that effort pays off.
“I don’t see the point.” Students invest effort in work they find interesting, useful, or connected to who they are or want to become. When an assignment feels arbitrary or disconnected from anything they care about, the effort of doing it themselves doesn’t feel worth it, and AI becomes the path of least resistance. Helping a student find a genuine reason the work matters to them will boost their motivation to dig into it.
“I’m running on empty.” Sometimes the issue isn’t the assignment; it’s everything around it. In a nationally representative survey, 81% of U.S. teens reported feeling intense pressure around achievement, their appearance, and their future. This pressure manifests itself in more than a quarter of teens saying they were struggling with burnout. Under these circumstances, Common Sense Media’s researchers note, reaching for AI can feel “less like cheating than what is needed to keep up.” When the perceived cost of doing the work (in time, stress, and sleep) climbs too high, offloading becomes a rational form of triage. Doing what you can to ease the pressure at home and protecting basics like sleep and downtime will prevent maladaptive uses of AI.
“If I have to struggle, I must not be good at this.” Students who believe ability is fixed, that you either have it or you don’t, tend to read struggle as proof they lack the talent, so they avoid situations that expose it. AI lets them sidestep this struggle and the uncomfortable self-judgment that comes with it. Students who instead see ability as something that grows with effort treat struggle as part of learning and therefore, embrace it. The Student Guide builds this “growth mindset” directly; at home, talking about difficulty and mistakes as valuable and important for learning will build your student’s growth mindset.
“It’s really just about the grade.” When school comes to feel like a contest over grades or test scores, students sensibly optimize for grades, and AI is often the fastest route to a good one. A student focused on learning asks, “Do I understand this?” whereas a student focused on performance asks, “How do I get this done and get the points?” The more messages around a student (including at home) emphasize outcomes (i.e., grades, test scores) over learning and growth, the more likely they are to use AI in ways that inhibit productive struggle.
Running underneath several of these is a simpler need: a sense of ownership. Students are more motivated when they feel some genuine say in their work and less motivated when everything feels imposed on them. AI, ironically, can give a disengaged student a small sense of control over a task they otherwise feel no ownership of.
Because these root causes respond to how the adults around a student respond, small changes at home can make a real difference. Appendix D gathers research-based moves for supporting your student’s motivation (Linnenbrink-Garcia et al., 2026).
The goal of addressing these root causes is not to eliminate AI use or pressure completely. It’s about addressing the reasons schoolwork feels skippable in the first place, so that productive struggle feels worth the effort. A motivated student with access to AI tends to use it as a tool, whereas a student struggling with motivation will be inclined to use AI as an escape hatch.
How Teachers Use AI, and What It Means for Your Student
Teachers are working out how to use AI in real time, much like the rest of us. It shows up in their work in a few broad ways, and knowing what they are helps you make sense of what your student experiences in class, and what to ask about when something seems off.
Michigan Virtual’s 2026 survey of Michigan educators found that teacher-reported classroom use of AI more than doubled between 2024 and 2026, and that more than 80% of educators now use AI in both their personal and professional lives (Michigan Virtual, 2026). But the same survey also found that even as use grew, educators’ trust in AI declined from the year before, and teachers and building leaders closest to the classroom reported more caution than district administrators did. Educators are generally adopting these tools while maintaining healthy skepticism, much as this guide encourages you and your students to do. The Michigan educators included in the survey echoed the same concerns featured in this guide: that AI could reduce students’ chances of thinking for themselves, doing authentic work, and replacing the human connection at the heart of school.
Behind the scenes. Most teacher AI uses never directly reach a student. Teachers use it most to draft worksheets and assessments, handle administrative work, and prepare lessons; in the Gallup-Walton survey, about 6 in 10 teachers reported using AI in some capacity. This is largely the time-saving benefit described earlier, and the payoff for your student is a teacher with more time for the human parts of teaching. The thing to know is that some of the materials or feedback your student receives may be AI-assisted, which is perfectly fine when a teacher reviews them and acts as the “human in the loop”.
For feedback and instruction. Some teachers go further, using AI to help grade, draft feedback, or run adaptive tutoring tools with students. Here, strategically maintaining the human-in-the-loop is critical: an AI-generated grade or comment should be checked by a person before your student sees it. If your student receives feedback or a grade that seems wrong or overly generic, it’s reasonable to ask whether and how the teacher reviewed it.
AI detection tools. One specific use deserves its own flag because the stakes for your student are high: many schools use tools that claim to detect whether writing was AI-generated. Many of these tools are unreliable and can be biased. A peer-reviewed benchmark study found that detectors falsely flagged about 61% of essays written by non-native English speakers as AI, while rarely misflagging native speakers, and later research finds neurodivergent students are also flagged at elevated rates (Liang et al., 2023). Newer detectors vary, and some vendors dispute these numbers, but it is important that educators use detectors with caution. A detector flag is a reason to look more closely and talk with the student, but it should not be used as proof. If your student is ever accused of using AI based on a single detector, they deserve a conversation and a chance to show their process, drafts, notes, and version history, rather than an automatic penalty.
Inconsistency is normal right now. Because teachers are adopting AI at different speeds and with little training, expectations vary widely, sometimes from one classroom to the next. In Common Sense Media’s 2026 survey, only about half of teens said their teachers have similar AI rules, and only 44% had ever discussed with a teacher when AI use is and isn’t allowed. The training gap is real on the adult side, too: Senate testimony citing a national survey noted that most schools provide little or no professional development on safe AI use to teachers (Mote, 2026). Michigan schools are moving from experimenting with AI toward formally building it into their plans; in that 2026 Michigan Virtual survey, a large majority of responding administrators said AI was already part of their district or school planning, up sharply from two years earlier, and in 2026 the Michigan Department of Education—together with Michigan Virtual—released statewide AI guidance for districts (Michigan Department of Education, 2026). Taken together, this means that AI is likely to become a more established and more openly discussed part of your student’s school experience, which makes this a good moment to start asking questions.
Access isn’t equal, and the classroom is where it shows. AI’s benefits and risks don’t reach every student the same way, and teacher and classroom use is often where the gaps become visible. Roughly one in six U.S. school-aged children still lack reliable home internet, and the shortfall falls hardest on low-income families, rural communities, students of color, English learners, and students with disabilities. When assignments require a device and a home connection, whether AI-assisted or not, students without them are pushed to rely entirely on school-issued devices or to fall behind. This divide runs the other way as well: for some students, the AI-powered accessibility tools described in the benefits section are a genuine lifeline, which is why well-meaning blanket bans can accidentally strip away accommodations those students depend on. As a parent, you should know that if home access is a barrier, your child’s school may be able to provide devices, hotspots, or connectivity assistance. If your student uses AI-based accommodations, make sure any classroom AI rules do not restrict them.
Your students’ teachers are navigating AI alongside you, often without much guidance of their own. That makes clear, two-way communication between home and school very valuable for navigating these new waters.
3 This is the flip side of the self-efficacy risk from the last section: low confidence drives AI use, and over-reliance then further erodes confidence, a loop worth catching early.
Where and How AI Shows Up Outside of School
School is only part of your student’s AI landscape. Away from the classroom, AI shows up in ways that are more personal, more emotional, and often harder for you to see: AI can function as a companion your student confides in, as an influence on their mental health, be embedded in toys for younger children, and as a tool for a serious new form of cyberbullying. Because so much of this happens on personal devices and in private chats, your awareness is just as important here as with schoolwork.
AI Companions and Relationships
Some of the most popular AI tools among young people aren’t for homework at all. AI companions are chatbots designed for conversation, friendship, and sometimes romance, built to feel like a relationship rather than a tool. They are already mainstream: Common Sense Media found that nearly three in four teens have used an AI companion, and about half use one regularly (Common Sense Media, 2025). The appeal is understandable. A companion is available at any hour, never judges, never gets bored, and, because of the agreeableness we discussed earlier, tends to tell a young person what they want to hear.
That design is exactly what makes companions risky for developing minds. Research from the University of Cambridge describes an “empathy gap”: children are especially likely to treat a chatbot as a lifelike, trusted confidante, while the chatbot itself handles the emotional, ambiguous parts of a real conversation poorly and can respond in confusing or unsafe ways that a child is less equipped to catch (Kurian, 2024). Kids also tend to disclose more to a friendly-seeming machine than to an adult. Meanwhile, an always-available, frictionless “friend” can crowd out the harder, messier practice of real relationships, the same human-connection concern raised in the risks section. In one 2025 survey cited in congressional testimony, a majority of schools reported that students confided in AI chatbots rather than teachers or parents, and many reported that students formed emotional attachments to them (Mote, 2026).
The expert consensus is notably firm. After testing popular companion products with mental-health specialists at Stanford, Common Sense Media concluded that social AI companions pose unacceptable risks and should not be used by anyone under 18 (Common Sense Media, 2025). What you can do: find out whether your student uses an AI companion, talk openly about the difference between an AI that simulates caring and a person who actually cares, and stay alert to signs of dependence, secrecy, or withdrawal from real-world friendships.
Mental Health Risks 4
Since AI companions are agreeable and always available, a struggling young person can fall into a loop where the AI reinforces their lowest moments and darkest feelings rather than interrupting them. In testing, researchers have documented companions engaging with or even encouraging dangerous behavior, including self-harm and other crises, instead of redirecting the young person toward help, and safety features that fail at exactly the wrong moment, as noted in the section on safety (Common Sense Media, 2025). There have been tragic real-world cases, including one widely reported death that led to a lawsuit, in which a teen in distress received harmful responses from a chatbot (Raine v. OpenAI, 2025). These cases are not everyday experiences, but they show what’s possible when a vulnerable young person leans on a tool that isn’t equipped to help and lacks sufficient safety guardrails.
If you’re ever concerned about your student’s safety or mental health, you don’t have to handle it alone. The 988 Suicide & Crisis Lifeline offers free support from trained counselors, 24/7; you or your student can call or text 988, or chat at 988lifeline.org. For non-crisis concerns, your pediatrician or a licensed mental-health professional can help you find the right support.
AI in Toys
AI is now being built into toys, plush animals and robots with embedded chatbots, marketed to young children as educational, screen-free companions. Two independent 2025–2026 investigations found serious problems. The U.S. PIRG Education Fund’s annual Trouble in Toyland report found AI toys that gave children unsafe guidance about dangerous household items and, in some cases, engaged with sexual content; after the report, one manufacturer pulled its product and the AI provider suspended the developer (U.S. PIRG Education Fund, 2025). Common Sense Media’s testing led it to rate AI companion toys an “unacceptable risk” for young children, finding that roughly a quarter of one toy’s responses were not age-appropriate, including content related to self-harm and other unsafe topics, and that the toys are designed to foster emotional attachment while collecting extensive data, including recordings of a child’s voice (Common Sense Media, 2026).
Young children are especially vulnerable here because, developmentally, they’re prone to “magical thinking” and may not grasp that the toy isn’t a real friend. In a companion survey, about half of parents of children aged 0–8 had bought or were considering an AI toy, but only about one in five actually wanted it to behave like a companion (Common Sense Media, 2026). Common Sense’s guidance for this age group is straightforward: it recommends avoiding AI toys for children age 5 and under and urges extreme caution for ages 6 to 12. For young children, traditional toys, books, and human interaction remain the safer, more developmentally appropriate choice. As a parent, you should approach AI toys for young kids with caution, check what data a toy collects and whether you can control it, and supervise use rather than treating the toy as a stand-in for human interaction.
Deepfakes and Cyberbullying
AI has made it easy to fabricate realistic images, video, and audio, and one use has hit schools hard: apps that generate fake nude images from ordinary photos. Students have used these to create and share explicit fakes of classmates, a form of harassment that can cause victims, overwhelmingly girls, lasting trauma. It’s more common than many parents realize. A Center for Democracy & Technology survey found that in a single school year, 40% of students and 29% of teachers were aware of deepfakes depicting people at their school, not all of them explicit (Center for Democracy & Technology, 2024). Documented incidents have led to student prosecutions and expulsions in several states.
The legal landscape has moved quickly in response. The federal TAKE IT DOWN Act, signed in 2025, makes it a crime to publish non-consensual intimate images, real or AI-generated, including of minors, and requires online platforms to remove reported images within 48 hours; at least half of states have passed their own laws as well, including Michigan, whose Protection from Intimate Deep Fakes Act took effect in 2025, criminalizing the creation and sharing of non-consensual intimate deepfakes and giving victims a way to sue (Michigan Advance, 2025). There are two things you can do to address deepfakes and cyberbullying. First, talk with your student about being a creator: they should know that making or forwarding these images, even “as a joke,” is harmful and illegal. Second, you can address the possibility that they could be a target, that if it happens, it is not their fault, and that help exists.
If your student is victimized, you can report the images to the platform, contact the school, and use the National Center for Missing & Exploited Children’s free Take It Down service (takeitdown.ncmec.org), which helps remove explicit images of minors from participating platforms.
4 This is a sensitive topic. The information here is meant to help you support your student; it isn’t a substitute for professional care.
Talk to Your Student About AI
The most powerful safeguard your student has isn’t a setting or a filter; it’s an ongoing conversation with you. This section gives you strategies and language for having conversations about AI both in and out of school. As a resource for you, we provide specific guidance tailored to five stages of development, with quick-reference tables you can return to in Appendix B (on AI for schoolwork and learning) and Appendix C (on AI outside of school).
How to get the most from these conversation guides
Meet your students where they actually are. Kids develop at different rates, and a grade band is a rough guide. Your student may be ready for a “middle school” conversation in fifth grade, or may need an “upper elementary” approach well into middle school. You know your child; trust that knowledge, and adapt for individual needs, including those of neurodivergent learners.
Start earlier than feels necessary. AI reaches children young, through voice assistants, recommendation feeds, and even toys, so the youngest bands aren’t just there for show. The goal in early years is to plant simple habits and language you’ll build on later.
The stages build on each other. Each age band assumes the foundations from the ones before it. As your student grows, you will build upon earlier conversations, add nuance to their thinking, increase independence, and expand access.
Aim for dialogue, not a lecture. The most useful AI conversations are two-way, curious, and repeated over time, not a single “talk.” Ask your student what they already know and think; you’ll learn a lot, and they’ll be far more open than if they feel they are being interrogated or criticized.
Keep one message constant at every age: “If something with AI ever feels confusing, wrong, or upsetting, you can come to me, and you won’t be in trouble for telling me.” Keeping the door open to your student matters more than any single rule.
Help Your Student Use AI Purposefully and Responsibly by Learning Together
Awareness and conversation lead somewhere concrete: to good AI use habits. Your job is not to become the family AI expert; it’s to be the steady, curious adult beside your student as you both figure this out. Researchers describe the ideal stance as that of co-learners: open, curious, and willing to navigate uncertainty together (UNICEF, n.d.), because right now, adults and kids are often learning at a similar pace.
Ask Your Student Questions
Start with curiosity, not interrogation. Asking what AI tools your student uses, what they find helpful, and what frustrates them will teach you more and keep them far more open than leading with a “don’t cheat with AI” lecture, which tends to make kids shut down (UNICEF, n.d.). These questions also surface something important: when students misuse AI, it often reflects disengagement or struggle with the work itself rather than simple access to a tool, and when they turn to chatbots for company, that can signal something about their social world. In other words, AI is a useful lens for noticing what your student actually needs, so you can address the root causes rather than the symptoms.
The Student Guide points students toward some of the biggest open questions in AI, and these make excellent conversation fuel at home: Should an artist’s work be used to train AI without permission? Should AI-generated content be labeled? How should we handle AI replacing jobs, or using AI to screen job applicants when it carries bias? Should companies disclose AI’s environmental costs? There are no settled answers, which is exactly why they’re good for helping your student form their own views, and for guiding the boundaries you set together.
Learn About AI Together
As a co-learner with your student, you don’t need to have answers ready. All you need is a willingness to explore.
Experiment like a scientist. The Student Guide encourages students to treat AI like an experiment: try a prompt, change it, notice what happens. Do this together. Type one of your student’s questions into a chatbot, look at the answer side by side, and talk it through: Which parts seem helpful? Which parts seem off or made up? How might the tool have produced that? This single experiment builds critical judgment better than a lecture would.
Notice how AI shapes the world, together. Spot it in the wild: the recommendations in a feed, an AI overview atop a search, a suspiciously perfect image. Reinforce the Student Guide’s line that “seeing isn’t always believing” now that convincing fakes are easy to make and increasingly difficult to detect.
Talk about ethics, responsibility, and risks. Use the debates above and the risks from earlier sections, bias, consent, and environmental cost, to connect AI use to your family’s values, not just rules.
Check before creating. Consider echoing a concrete rule from the Student Guide: get someone’s permission before using their photo, writing, or voice in an AI tool; never create deepfakes or content that misrepresents what someone said or did; and always respect it when someone says no. This is a way to get the deepfake conversation started.
Make WISE Decisions Together
The Student Guide gives students a simple gut-check to run before they use AI, the WISE framework (Quidwai, 2024), and you can use the same four questions at home as a shared decision tool:
W — Wellbeing: How will using AI here affect my wellbeing, and other people’s?
I — Integrity: Am I using AI honestly and fairly?
S — Skills: What will I learn by using it, and what might I miss learning?
E — Engage: Does using AI help me engage more deeply with what I’m doing, or less?
Because your student is already learning this framework, using its language at home gives your family a shared, low-friction way to talk through a specific decision without it becoming a fight.
Set Family Boundaries
There’s no universal rulebook, which means your family gets to decide what feels right. Talk through practical questions together: which tools are okay and for what, when and where they should be used (e.g., for younger children, keeping devices in shared spaces helps), what information should never be typed in, when AI use should be disclosed and how, and what times or activities stay AI-free. Experts recommend pairing simple, agreed-upon boundaries with regular check-ins (UNICEF, n.d.). Expect these to evolve; as your student grows and the tools change, the boundaries should too. The “Questions to Ask About an AI Tool” checklist in Appendix A is a good thing to run through together before adopting something new.
Notice Your Student’s AI Use
It is important to stay aware of how your student is using AI. There are a few warning signs worth watching for: heavy use or real distress when asked to stop; behavioral changes like becoming secretive or anxious, or leaning on AI for emotional support instead of people. If you notice these, lead with gentle, open questions rather than criticism, asking what your student likes about the tool and whether anything about it feels off. And remember the throughline from the motivation section: a pattern of unhealthy use is usually a symptom. Addressing the underlying needs, pressures, disengagement, and loneliness will be more effective than policing behavior.
Model Good AI Habits
Your student is watching how you use AI. Modeling does quiet, powerful work: let them see you double-check an AI answer, decline to paste private or others’ information into a tool, admit when AI got something wrong or when you’re unsure how it works, and keep your own healthy limits around screens and tools. Narrating your own choices out loud (“I’m going to verify this before I trust it”) turns everyday moments into lessons and reinforces the co-learner spirit far better than rules you don’t follow yourself.
Talk to Your Student’s Teacher(s) About AI
Because school AI rules vary so much, from class to class and school to school, asking directly is one of the most useful things you can do. Ask how the school approaches AI: how it ensures AI supports learning, how it handles academic integrity and disclosure, and how it thinks about student wellbeing. It’s also worth asking about the school’s broader philosophy of learning, since that context matters as much as any AI policy or guidance (UNICEF, n.d.). If your student uses a specific classroom AI tool, the vendor questions from earlier in this guide are fair to raise with the school, too.
A Final Word
You don’t have to master this new arrival technology to be a good parent. AI is important, but it isn’t the whole story. What shapes your student most isn’t any single tool; it’s their relationships, routines, interests, and the people who support them, and you are the center of that. Stay curious, keep the conversation open, and hold onto the one message that matters at every age and in every section of this guide: whatever happens with AI, your student can come to you. That open door is the most powerful safeguard there is.
References
The AI Education Project (aiEDU). (2024). AI literacy framework. https://www.aiedu.org/
Bandura, A. (1997). Self-efficacy: The exercise of control. W. H. Freeman.
Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences, 122(26). https://doi.org/10.1073/pnas.2422633122
Burns, M., Winthrop, R., Luther, N., Venetis, E., & Karim, R. (2026). A new direction for students in an AI world: Prosper, prepare, protect. Brookings Institution. https://www.brookings.edu/articles/a-new-direction-for-students-in-an-ai-world-prosper-prepare-protect/
Center for Democracy & Technology. (2024). In deep trouble: Surfacing tech-powered sexual harassment in K-12 schools. https://cdt.org/insights/report-in-deep-trouble-surfacing-tech-powered-sexual-harassment-in-k-12-schools/
Cheng, M., Lee, C., Khadpe, P., Yu, S., Han, D., & Jurafsky, D. (2026). Sycophantic AI decreases prosocial intentions and promotes dependence. Science, 391(6792), eaec8352. https://doi.org/10.1126/science.aec8352
Common Sense Media, Youth AI Safety Institute. (2025). Social AI companions risk assessment. https://institute.commonsensemedia.org/risk-assessments/social-ai-companions
Common Sense Media, Youth AI Safety Institute. (2026). AI risk assessment: AI toys. https://www.commonsensemedia.org/ai-ratings/ai-toys
Common Sense Media. (2024). The dawn of the AI era: Teens, parents, and the adoption of generative AI at home and school. https://www.commonsensemedia.org/research/the-dawn-of-the-ai-era-teens-parents-and-the-adoption-of-generative-ai-at-home-and-school
Common Sense Media. (2025). Meta AI companions unsafe for kids, Common Sense Media report finds [Press release]. https://www.commonsensemedia.org/press-releases/meta-ai-companions-unsafe-for-kids-common-sense-media-report-finds
Common Sense Media. (2025). Talk, trust, and trade-offs: How and why teens use AI companions. https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions
Common Sense Media. (2026). Teens in the AI era: Schoolwork and skills that matter (Zimmermann, L., Mann, S., Radesky, J., & Robb, M. B.). https://www.commonsensemedia.org/sites/default/files/research/report/commonsensemedia_teensaischoolworkskills_2026.pdf
Common Sense Media. (n.d.). AI ratings and reviews. https://www.commonsensemedia.org/ai
Dweck, C. S. (2006). Mindset: The new psychology of success. Random House.
Eccles, J. S., & Wigfield, A. (2020). From expectancy-value theory to situated expectancy-value theory: A developmental, social cognitive, and sociocultural perspective on motivation. Contemporary Educational Psychology, 61, 101859. https://doi.org/10.1016/j.cedpsych.2020.101859
The Education Trust–West. (2024). Education equity in crisis: The digital divide. https://west.edtrust.org/resource/education-equity-in-crisis-the-digital-divide/
Emmons, S. (2026). AI-powered chatbots can become too agreeable over time, researchers report. Penn State News. https://www.psu.edu/news/information-sciences-and-technology/story/ai-powered-chatbots-can-become-too-agreeable-over-time
eSafety Commissioner. (2025). Social media age restrictions. Australian Government. https://www.esafety.gov.au/about-us/industry-regulation/social-media-age-restrictions
Federal Trade Commission. (2025, April 22). Children’s Online Privacy Protection Rule [Final rule]. Federal Register. https://www.federalregister.gov/documents/2025/04/22/2025-05904/childrens-online-privacy-protection-rule
Federal Trade Commission. (n.d.). Protecting your child’s privacy online. Consumer Advice. https://consumer.ftc.gov/articles/protecting-your-childs-privacy-online
Fesler, L., Martinez Claeys, J., Agnew, C., & Loeb, S. (2026). The evidence base on AI in K-12: A 2026 review. Stanford SCALE Initiative. https://scale.stanford.edu/sites/default/files/The%20Evidence%20Base%20on%20AI%20in%20K-12%20Report.pdf
Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006
Grinschgl, S., Papenmeier, F., & Meyerhoff, H. S. (2021). Consequences of cognitive offloading: Boosting performance but diminishing memory. Quarterly Journal of Experimental Psychology, 74(9), 1477–1496. https://doi.org/10.1177/17470218211008060
Han, S., et al. (2025). Children’s mental models of AI reasoning: Implications for AI literacy education. Proceedings of the 24th Interaction Design and Children Conference (IDC ’25). https://dl.acm.org/doi/10.1145/3713043.3728856
Internet Safety Labs. (n.d.). Product safety audits. https://internetsafetylabs.org/
Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X.-H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task [Preprint]. arXiv. https://arxiv.org/abs/2506.08872
Kurian, N. (2024). “No, Alexa, no!”: Designing child-safe AI and protecting children from the risks of the “empathy gap” in large language models. Learning, Media and Technology, 621–634. https://doi.org/10.1080/17439884.2024.2367052
The Learning Agency & Renaissance Philanthropy. (2026). Guidelines for the safe and responsible use of AI in ed tech. https://the-learning-agency.com/guides-resources/guidelines-for-the-safe-and-responsible-use-of-ai-in-ed-tech/
The Learning Agency. (2026). Closing the child speech recognition gap: Evidence, limitations, and paths forward. https://the-learning-agency.com/guides-resources/closing-the-child-speech-recognition-gap-evidence-limitations-and-paths-forward/
LearnLM Team, Google. (2025). AI tutoring can safely and effectively support students: An exploratory RCT in UK classrooms [Preprint]. arXiv. https://arxiv.org/abs/2512.23633
LearnPlatform by Instructure. (2025). EdTech Top 40: K-12 edtech engagement during the 2024–25 school year. https://www.instructure.com/press-release/new-learnplatform-instructure-report-shows-k-12-districts-are-more-selective-about
Lee, V. R., & Pope, D. (2024). Cheating in the age of generative AI: A high school survey study of cheating behaviors before and after the release of ChatGPT. Computers and Education: Artificial Intelligence. https://www.sciencedirect.com/science/article/pii/S2666920X24000560
Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7), 100779. https://doi.org/10.1016/j.patter.2023.100779
Linnenbrink-Garcia, L., Snyder, K. E., Perez, T., Wormington, S. V., Riggsbee, J., Ben-Eliyahu, A., & Hill, N. E. (2018). Multiple pathways to success: An examination of integrative motivational profiles among upper elementary and college students. Journal of Educational Psychology, 110(7), 1026–1048. https://doi.org/10.1037/edu0000245
Linnenbrink-Garcia, L., Wigfield, A., Cabrera, L., Mouzaoir, S., McKinney, D., Liu, P. P., Marchand, G., Schmidt, J. A., Harris, C. J., & Guthrie, J. (2026). The case for multifaceted, integrative motivation interventions embedded in classrooms. Educational Psychology Review, 38(1), 89. https://doi.org/10.1007/s10648-026-10177-w
Michigan Advance. (2025, August 27). Whitmer approves ban on non-consensual deep fake pornography. https://michiganadvance.com/briefs/whitmer-approves-ban-on-non-consensual-deep-fake-pornography/
Michigan Department of Education. (2026). Artificial intelligence [Guidance resources]. https://www.michigan.gov/mde/services/academic-standards/educational-technology/artificial-intelligence
Michigan Department of Education. (n.d.). Pupil privacy. https://www.michigan.gov/mde/pupil-privacy
Michigan Virtual. (2026). AI in education: A 2026 snapshot of growing use and the shift toward integration. https://michiganvirtual.org/research/publications/ai-in-education-a-2026-snapshot-of-growing-use-and-the-shift-toward-integration
Michigan Virtual. (n.d.). Student guide to AI. https://michiganvirtual.org/resources/guides/student-guide-ai
Midgley, C. (Ed.). (2002). Goals, goal structures, and patterns of adaptive learning. Lawrence Erlbaum Associates.
Mote, E. (2026, June 16). Written testimony: Hearing on the future of K-12 education in the age of artificial intelligence [Testimony]. U.S. Senate Committee on Health, Education, Labor, and Pensions. https://www.help.senate.gov/imo/media/doc/cf1d05e5-ac56-a498-0ee6-8a9b641b540c/Mote%20Testimony_431af40b-3932-4e62-9bb5-2787c2393522.pdf
Mozilla Foundation. (n.d.). Privacy not included: A buyer’s guide for connected products. https://foundation.mozilla.org/privacynotincluded/
National Institute of Standards and Technology. (2023). Artificial intelligence risk management framework (AI RMF 1.0). U.S. Department of Commerce. https://doi.org/10.6028/NIST.AI.100-1
OECD & European Commission. (2025). Empowering learners for the age of AI: An AI literacy framework for primary and secondary education. OECD Publishing.
Pew Research Center. (2025, January 15). About a quarter of U.S. teens have used ChatGPT for schoolwork — double the share in 2023. https://www.pewresearch.org/short-reads/2025/01/15/about-a-quarter-of-us-teens-have-used-chatgpt-for-schoolwork-double-the-share-in-2023/
Pew Research Center. (2026, February 24). How teens use and view AI. https://www.pewresearch.org/internet/2026/02/24/how-teens-use-and-view-ai/
Public Interest Privacy Center. (2024). The K-12 privacy policy guide: How to quickly spot red flags. https://publicinterestprivacy.org/
Quidwai, S. (2024). Designing schools: How design thinking makes you irreplaceable in the age of AI.
Raine v. OpenAI, No. CGC-25-628528 (Cal. Super. Ct. filed Aug. 26, 2025).
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68–78. https://doi.org/10.1037/0003-066X.55.1.68
Student Online Personal Protection Act, Public Act 368 of 2016 (Michigan). https://www.legislature.mi.gov/documents/mcl/archive/2025/August/mcl-Act-368-of-2016.pdf
U.S. Department of Education. (n.d.). A parent guide to the Family Educational Rights and Privacy Act (FERPA). Student Privacy Policy Office. https://studentprivacy.ed.gov/resources/parent-guide-family-educational-rights-and-privacy-act-ferpa
U.S. PIRG Education Fund. (2025). Trouble in Toyland 2025: AI bots and toxics present hidden dangers. https://pirg.org/edfund/resources/trouble-in-toyland-2025-a-i-bots-and-toxics-represent-hidden-dangers/
UNESCO. (2021). Recommendation on the ethics of artificial intelligence. https://www.unesco.org/en/artificial-intelligence/recommendation-ethics
UNESCO. (2024). Challenging systematic prejudices: An investigation into bias against women and girls in large language models. https://unesdoc.unesco.org/ark:/48223/pf0000388971
UNICEF. (n.d.). Parenting in the AI age: How to approach AI with your children (interview with Y. Xu). https://www.unicef.org/parenting/digital-parenting/how-approach-ai-children
Walton Family Foundation & Gallup. (2025). Teaching for tomorrow: Unlocking six weeks a year with AI. https://news.gallup.com/poll/691967/three-teachers-weekly-saving-six-weeks-year.aspx
Zhang, S., & Xu, J. (2024). The paradox of self-efficacy and technological dependence: Unraveling generative AI’s impact on university students’ task completion. The Internet and Higher Education. https://www.sciencedirect.com/science/article/abs/pii/S109675162400040X
Appendices: A Parent Toolkit
You don’t need technical expertise to size up an AI tool. These questions, adapted from guidance for the people who build and buy ed tech, work whether you’re checking an app your student downloaded or one their school assigned.Where to check independently: Common Sense Media (privacy ratings and product risk assessments), the nonprofit Internet Safety Labs, Mozilla’s Privacy Not Included buyer’s guide, and the Public Interest Privacy Center’s “K-12 Privacy Policy Guide: How to Quickly Spot Red Flags.” You can also ask your student’s school which tools they’ve formally vetted (Common Sense Media, n.d.; Internet Safety Labs, n.d.; Mozilla Foundation, n.d.; Public Interest Privacy Center, 2024).
Area | Questions to ask |
Does it help my student learn? | Does it coach my student toward an answer, or just hand one over? Is there any evidence that it actually improves learning? |
What happens to our data? | What information does it collect, and why? Is my student’s data used to train the company’s AI, sold, or used for ads? Can we review, export, or delete it? |
Is it safe? | What kinds of content does it block? How does it respond if my student expresses distress? Does it make clear that it’s AI, not a person? |
Can I trust what it says? | Where does it get its answers, and does it cite sources or admit when it’s unsure? Does a human review AI-generated grades or feedback before students see them? |
Is it upfront? | Are its data practices written in plain language you can find and read? Who do you contact with a question or concern? |
The core message stays the same at every age: AI should help your student learn, not learn for them.
Stage | What’s happening developmentally | What to focus on | Try saying |
Early childhood (pre-K) | Prone to “magical thinking”; treat talking devices as alive; anthropomorphize responsive tech more than older kids. | A talking device is a machine, not a person who “knows” things; figuring things out yourself is fun. Don’t let a device be the default answer-giver. | “That’s a computer talking, not a person, and computers can be wrong.” “Good question! What do you think the answer is first?” |
Early elementary (K–2) | Foundational-skill years; concrete thinkers who trust authority and take on-screen answers at face value. | AI can be wrong, so we check it; the point of practice is to grow your own brain, which takes effort. | “Computers make mistakes just like people do, so how could we check that?” “Your brain grows when you do the hard part, kind of like a muscle.” |
Upper elementary (3–5) | More independent schoolwork; developing metacognition; can grasp using AI for some things but not others. | The difference between AI that helps you learn and AI that does the learning for you; try it yourself first, then use AI to check or improve. | “Did the AI help you understand it, or did it just hand you the answer?” “Try it on your own first, then ask AI, and see whether you agree.” |
Middle school (6–8) | Abstract reasoning emerges alongside peer influence and early grade pressure; AI rules feel unclear. | What counts as help vs. cheating (and that it varies by class, so ask); why struggle matters; pushing back on AI’s agreement. | “What’s your teacher’s rule for AI on this one?” “AI usually agrees with you even when you’re off base—did you check it?” “If AI does the thinking now, what happens when it’s just you on the test?” |
High school (9–12) | Capable of ethical reasoning; high stakes and real pressure; heavy AI use is the norm. | Their own line between tool and crutch; honest disclosure; building vs. borrowing skills; the pressure underneath. | “Where’s your line between AI helping you and AI doing it for you?” “Are you building that ability or renting it?” “Are you using AI because it helps you learn, or because you’re slammed?” |
Outside the classroom the risks are more personal. The steady thread: your student should always know they can come to you if something feels wrong, and won’t be in trouble for telling you.
Stage | What’s happening developmentally | What to focus on | Try saying |
Early childhood (pre-K) | Anthropomorphize readily; can’t reliably separate real from pretend; most likely to bond with a “companion” toy. | Decisions are yours: lean on traditional toys, books, and real interaction; supervise any AI toy; reinforce that the toy is a machine. | “Your bear can talk, but it’s a toy, not a real friend, and it doesn’t really have feelings.” “Thank you for telling me—let’s put that one away.” |
Early elementary (K–2) | Still treat responsive tech as humanlike and trustworthy; beginning to use voice assistants and simple chatbots. | A chatbot can sound like a friend, but it is a computer that can’t truly care or keep them safe; we don’t tell devices private things. | “A chatbot can be fun, but it’s a computer, not a friend. If it says something weird or scary, come tell me.” “We keep private things, like where we live, to ourselves.” |
Upper elementary (3–5) | Developing more selective trust, spending more time online, may start experimenting with chatbots, and can grasp that images and videos can be faked. | AI can create realistic fakes, so “seeing isn’t always believing”; an AI “friend” isn’t a real relationship; early kindness, online habits. | “AI can make fake photos and videos that look totally real, so we can’t always trust what we see.” “An AI chatbot can be fun, but it can’t actually care about you the way real people do.” |
Middle school (6–8) | Identity, belonging, and peer approval are central; emotion regulation is still developing; the critical age for the deepfake conversation. | Companions can’t replace real people (their “empathy” is simulated); AI is not a counselor; deepfakes and cyberbullying are harmful and illegal—and never the victim’s fault. | “It can feel easier to tell things to an AI, but it’s no substitute for real people.” “If you’re struggling, come to me or another adult—or call or text 988.” “Making or forwarding fake nude or embarrassing pictures, even as a joke, is illegal and hurts people.” |
High school (9–12) | Can reason about ethics and consent; forming intense relationships (sometimes with AI); highest exposure to deepfakes and sextortion, with the most privacy from you. | What an AI can and can’t give; an agreeable AI can reinforce dark thoughts; consent and stepping in (not spreading); help and removal, not shame. | “What can an AI companion actually give someone, and what can’t it?” “I’d rather you talk to me or a professional than an app that can get it wrong; 988 is there anytime.” “If you’re ever targeted, come to me—no judgment. We can report it and get it removed.” |
If explicit images of your student (real or AI-made) are ever shared, the National Center for Missing & Exploited Children’s free Take It Down service (takeitdown.ncmec.org) can help remove them.
When a student routinely hands work to AI in a way that skips the learning, it usually signals an unmet motivational need rather than laziness. Listen for what’s underneath, and respond to the cause.
What you might hear | What it may signal | What helps |
“I don’t think I can do this.” | Low self-efficacy — doubt they can succeed. | Break tasks into pieces they can finish; help them string together small, real wins; skip empty “you’re so smart.” |
“I don’t see the point.” | Low perceived value. | Connect the work to something they care about, or an identity they’re growing into. |
“I’m running on empty.” | Burnout / high cost in time, stress, sleep. | Ease pressure at home; protect sleep and downtime. |
“If I struggle, I must not be good at this.” | Fixed mindset. | Treat difficulty and mistakes as normal and valuable; build a growth mindset. |
“It’s really just about the grade.” | Performance focus over learning. | Ask “what did you figure out?” more than “what did you get?”; ease grades-at-all-costs pressure. |
A few research-based moves that support motivation across the board:
Build confidence through real wins, not praise. Confidence comes from succeeding at something hard, not from being told they’re smart.
Help them find the “why.” A student who sees the point is far more willing to do the thinking.
Praise effort and strategy, and normalize struggle. “You kept working even when it was hard” builds resilience.
Give them some ownership. Offer real choices; ask more than you dictate. Ownership fuels motivation; control drains it.
Ask about learning, not just grades.
Be the available alternative to AI. Most teens try to solve a problem themselves or ask a person before turning to AI — so make your availability to help explicit, so asking a human feels easier than asking a chatbot.
(Bandura, 1997; Common Sense Media, 2026; Linnenbrink-Garcia et al., 2026)
You don’t need to be a privacy lawyer. Here is what the main laws do and what they mean for you.
Law | Who or what it protects | What it means for you |
COPPA — Children’s Online Privacy Protection Act | Children under 13; personal information collected online. | Companies need a parent’s consent before collecting a child’s personal information; you can review it, have it deleted, and stop further collection. A 2025 FTC update added biometric identifiers (such as voiceprints and face scans) and tightened data-sharing for advertising (Federal Trade Commission, 2025). |
FERPA — Family Educational Rights and Privacy Act | Student education records at federally funded schools (nearly all). | You can inspect records, request corrections, and control how they’re shared (rights transfer to the student at 18). A school’s AI vendor must use data only for educational purposes — not to train its own models or sell them. A school may consent to educational tools for under-13s, but not to commercial uses such as advertising (U.S. Department of Education, n.d.). |
Michigan Student Online Personal Protection Act (Public Act 368 of 2016) | Online tools and services students use for school. | Bars operators from selling students’ personal data, using it for targeted ads, or building non-educational profiles, and requires deletion at the school’s request. A separate provision (MCL 380.1136) bars schools and their contractors from selling education-record data to for-profit companies (Student Online Personal Protection Act, 2016; Michigan Department of Education, n.d.). |
To evaluate a specific tool against these protections, use the checklist in Appendix A.
988 Suicide & Crisis Lifeline — call or text 988, or chat at 988lifeline.org (free, 24/7 support from trained counselors).
NCMEC Take It Down — takeitdown.ncmec.org (free service to help remove explicit images of minors).
Common Sense Media — privacy ratings and AI product risk assessments for families (commonsensemedia.org/ai).
Public Interest Privacy Center — The K-12 Privacy Policy Guide: How to Quickly Spot Red Flags (publicinterestprivacy.org).
Internet Safety Labs and Mozilla’s Privacy Not Included — independent product and app safety and privacy reviews.
Your student’s school — ask which AI tools are approved, what data they collect, and how to opt out.