Your Student's School Is Running AI. Do You Know Which Kind?
A parent in Walnut Creek recently asked her district's tech coordinator a simple question: what AI tools are active in her ninth-grader's classes? She got a three-paragraph non-answer about "innovative learning environments" and a link to the district's general technology policy, last updated in 2019.
That story is not unusual. It is the norm.
Schools have deployed AI tools at scale over the past two years, and most parent-facing communication has ranged from vague to nonexistent. The reason this matters more than most tech rollouts is that the tools in question are not interchangeable. There are two distinct categories operating in classrooms right now, and they carry fundamentally different implications for your student.
The Split: Learning AI vs. Surveillance AI
These two categories get collapsed into a single "AI in schools" debate constantly. They should not be.
Learning AI is designed to interact with academic content. It adapts difficulty levels based on student performance, provides tutoring-style feedback, flags knowledge gaps for teachers, and personalizes pacing. The data it collects is primarily academic: quiz scores, response times, error patterns, engagement with specific content.
Surveillance AI is designed to monitor behavior, attention, and compliance. It can analyze facial expressions during video calls, track eye movement to detect "off-task" behavior, flag keystrokes for review, score emotional states, or generate risk profiles based on behavioral patterns across a school day.
Here is the side-by-side reality:
| Feature | Learning AI | Surveillance AI |
|---|---|---|
| Primary function | Personalize academic content | Monitor student behavior/attention |
| Data collected | Test scores, response patterns, learning pace | Facial expressions, keystrokes, behavioral flags, emotional scores |
| Who sees the output | Teacher, student, parent | Administrator, third-party vendor, possibly law enforcement |
| Student aware it's running? | Usually | Often not |
| Data retention | Typically tied to enrollment | Often indefinite, held by vendor |
| Regulated under FERPA? | Yes, with exceptions | Complicated, vendor contracts vary widely |
Examples of learning AI in practice: adaptive math platforms that adjust problem sets in real time, AI writing coaches that give sentence-level feedback, reading tools that adjust text complexity.
Examples of surveillance AI in practice: proctoring software that records webcam footage and flags suspicious movement, attention-tracking tools that score how long a student appears to look at the screen, behavioral prediction systems that assign risk scores based on aggregated school activity.
Both categories exist. Both are active in U.S. schools. Many parents cannot tell you which one their student encountered today.
The Adoption Snapshot
Hard, verified numbers on AI deployment rates in K-12 schools are genuinely scarce, most surveys capture intent to adopt, not confirmed active use. Research organizations like RAND and the EdWeek Research Center have ongoing work in this space, but district-level transparency is inconsistent enough that self-reported data has real limits.
What is documentable: the market for both categories has grown sharply since 2022. Online proctoring, which sits squarely in the surveillance category, expanded rapidly during pandemic-era remote learning and did not fully contract when schools returned to buildings. Adaptive learning platforms have proliferated across math and literacy instruction at both the K-8 and high school levels.
Families already operating outside the traditional system, including those at programs like Rocklin Online High School or Richardson Online High School, often have more direct visibility into what tools their student's platform uses because the vendor relationship is more transparent by design. In large district settings, that visibility is frequently absent.
The Data Risk: What's Actually at Stake
FERPA (the Family Educational Rights and Privacy Act) gives parents rights over education records. The complication is that AI tool vendors often argue the data they collect does not qualify as an education record, it is "operational data" or "product improvement data", which can place it outside FERPA's protections entirely.
Three specific risks deserve attention:
Third-party data sharing. When a school licenses an AI tool, the vendor's terms of service govern what happens to the data. Some vendors explicitly prohibit resale of student data. Others permit use for "product improvement," which can mean training AI models on your student's responses, behavior patterns, and flagged content.
Breach exposure. Student data is a high-value target. Educational institutions have appeared in breach disclosures repeatedly in recent years, and third-party vendors compound the exposure surface. When a vendor suffers a breach, the school's own security posture is irrelevant.
Behavioral scoring and permanence. Surveillance AI outputs, including risk scores and behavioral flags, can persist in student records in ways that follow students forward. A score generated by an algorithm in seventh grade is not inherently accurate, and its downstream effects on disciplinary records or academic placement are not always visible to parents.
State law adds another layer of inconsistency. California has relatively strong student privacy statute. Other states have minimal protections beyond federal baseline. The patchwork is real.
State Policy Snapshot (Generalized Tiers):
- Stronger state protections (explicit AI/ed-tech privacy statutes, data deletion requirements, vendor contract mandates): California, New York, Colorado, Illinois
- Moderate protections (general student privacy laws, limited AI-specific provisions): Texas, Florida, Ohio, Washington
- Minimal AI-specific policy (relying primarily on federal baseline): Most remaining states
If you are in New York, whether in Brooklyn, Poughkeepsie, or Schenectady, your state has enacted Education Law 2-d, which includes specific vendor data security requirements. That does not guarantee compliance. It means you have a legal hook if something goes wrong.
The Parent Checklist: 10 Questions That Demand Real Answers
Bring these to your principal, technology director, or school board. Vague answers are data points.
1. What AI tools are currently active in my student's classes?
You want a list. Not a category. Not a philosophy statement. A list.
2. For each tool: is it learning-focused or does it monitor behavior or attention?
If the administrator cannot answer this, they do not know what they deployed.
3. Where is my student's data stored, and who has access to it?
The answer should name a specific vendor and a specific data location. "The cloud" is not an answer.
4. Does the vendor's contract prohibit resale or third-party sharing of student data?
Ask to see the relevant contract clause. Schools are required to maintain these agreements.
5. What data does each tool collect beyond academic performance?
Push specifically on behavioral flags, facial analysis, keystroke logging, and emotional scoring.
6. How long is student data retained after my student leaves the school?
Perpetual retention by a vendor is a red flag. Ask for the deletion policy.
7. Has the school conducted a privacy impact assessment for each AI tool?
Many haven't. The question itself signals that you know what one is.
8. Can I opt my student out of any surveillance-category tools without academic penalty?
The answer to this question tells you a great deal about how the school views student rights.
9. Has any tool the school uses been involved in a data breach in the past three years?
This is public record in many cases. If the school doesn't know, that's the answer.
10. Who at the district level is responsible for AI tool vetting and oversight?
If there is no named person, there is no real oversight.
Red flags by tool type:
- Proctoring software that runs webcam analysis: ask specifically whether facial expression or eye-tracking data is retained.
- "Behavioral analytics" platforms: ask what triggers a flag, who reviews flags, and whether flags enter the student's record.
- Any tool described as monitoring "engagement" or "attention": this language almost always means passive surveillance is running.
What You Can Do Right Now
The families with the most leverage are the ones who ask first and loudest. Schools respond to informed pressure faster than they respond to general concern.
If your district cannot give you straight answers, your state's department of education likely has a student privacy office. Several states have student data privacy coalitions that track vendor compliance. Those are your escalation paths.
And if the answer to "what AI is running on my student's device" turns out to be "more than you're comfortable with," you have options. Families in Port Charlotte, Tyler, and Folsom have found that online homeschool programs offer a different relationship with technology: one where the parent, not the district, controls the terms. The comparison between those models is worth understanding. This breakdown of online homeschool vs. public virtual school is a direct starting point.
The tools your student uses to learn should be visible to you. The tools watching your student should require your explicit consent. Right now, neither of those standards is universal. Asking the ten questions above is how you find out exactly where your school stands.