In December 2024, a 19-year-old college student named Matthew Lane accessed PowerSchool's systems and exfiltrated years of student and staff records from school districts across the country. PowerSchool serves over 18,000 schools supporting approximately 60 million students across North America. The stolen data included Social Security numbers, birthdates, contact information, and legacy files dating back to 1985.
Lane wasn't a sophisticated nation-state actor. He was a student at Assumption University in Worcester, Massachusetts, who gained access via a compromised employee password to a system that lacked multifactor authentication. He remained undetected for nine days. PowerSchool paid a ransom, hoping the data would be deleted. By May 2025, cybercriminals were sending extortion emails to school districts using samples of the same stolen data.
The PowerSchool breach illustrates a painful reality: education is now one of the most targeted sectors in cybersecurity. According to the FBI's Internet Crime Complaint Center, educational institutions experienced a 30% increase in reported cyber incidents between 2022 and 2023. Schools depend on hundreds of ed-tech tools to manage sensitive student records, yet FERPA lacks clear cybersecurity requirements for the vendors handling that data.
Against this backdrop, schools are rapidly adopting AI tools. Teachers use ChatGPT to create lesson plans. Administrators use Copilot to draft communications. Counselors consider AI for summarizing student records. Each use case potentially exposes FERPA-protected information to systems that may not meet the law's requirements.
The short version: If you need to redact sensitive documents before they reach AI systems, PaperVeil handles that layer. The rest of this article explains where it fits in the broader governance architecture.
What FERPA Requires
The Family Educational Rights and Privacy Act protects the privacy of student education records. Understanding its requirements is essential before using any AI tool with student data.
Protected Education Records
FERPA protects "education records," defined as records directly related to a student that are maintained by an educational institution or a party acting on its behalf. This includes:
Academic records: Grades, transcripts, course enrollments, test scores, class schedules, and academic progress reports.
Disciplinary records: Behavioral incidents, suspensions, expulsions, and conduct violations.
Attendance records: Absence patterns, tardiness records, and enrollment status.
Special education records: Individualized Education Programs (IEPs), 504 plans, disability documentation, and accommodation records.
Health records: When maintained by the school rather than a healthcare provider, including immunization records, nurse visit logs, and health screenings.
Financial records: Free and reduced lunch applications, family financial information, and scholarship applications.
Counseling records: Guidance counselor notes, college application materials, and career planning documentation.
Parental and Student Rights
FERPA provides parents the right to inspect and review their children's education records, request amendments to records they believe are inaccurate, and provide consent before the school discloses personally identifiable information from education records.
When a student turns 18 or enrolls in a postsecondary institution at any age, they become an "eligible student" and all rights transfer from parent to student.
Disclosure Restrictions
Schools generally cannot disclose personally identifiable information (PII) from education records without prior written consent. FERPA defines PII broadly to include direct identifiers like names, addresses, and Social Security numbers, as well as indirect identifiers that could be used, alone or in combination, to identify a student.
The "school official" exception allows disclosure to school officials with legitimate educational interest, including contractors performing services that the school would otherwise perform itself. This exception is how schools authorize vendors to access student data. But it comes with requirements: the vendor must be under direct control of the school regarding data use and can only use the data for authorized purposes.
Directory Information
Schools may designate certain information as "directory information" that can be disclosed without consent. This typically includes names, addresses, phone numbers, enrollment status, and participation in activities. However, parents must be notified of what constitutes directory information and given the opportunity to opt out.
Why AI Creates FERPA Exposure
AI tools introduce new compliance challenges that FERPA wasn't designed to address.
The Third-Party Vendor Problem
When a teacher pastes student information into ChatGPT or Claude, that data flows to a third-party provider. For that disclosure to be permissible under FERPA's school official exception, several conditions must be met:
- The AI provider must be performing a service that the school would otherwise perform itself
- The school must maintain direct control over the provider's use of the data
- The provider must be prohibited from using the data for any purpose other than the authorized service
- The provider must meet the school's criteria for access to education records
Consumer AI products generally don't meet these requirements. Terms of service may allow data to be used for model training, retained indefinitely, or shared with subprocessors. The school doesn't maintain direct control. The data use may exceed authorized purposes.
The Training Data Risk
Some AI models store and use interactions for training. When student PII flows through these systems, it may become embedded in the model's training data, creating an unauthorized retention and use of education records.
Even AI providers that offer "no training" options may retain conversation logs for abuse detection, quality assurance, or compliance purposes. These retention practices must be evaluated against FERPA's disclosure restrictions.
The Re-identification Challenge
FERPA allows disclosure of de-identified data, but sets a high bar: information that has been "recorded in such a manner that the identity of the student is not personally identifiable through single or multiple releases, and taking into account other reasonably available information."
AI systems are exceptionally good at re-identification. Studies have shown that AI can re-identify individuals from seemingly anonymous data by combining multiple indirect identifiers. The more data points included, the higher the re-identification risk. FERPA's de-identification standard may be harder to meet than many schools realize when AI is involved.
The Shadow AI Problem
Teachers and administrators under productivity pressure will use tools that help them work faster. If approved AI workflows aren't available or convenient, staff may turn to personal accounts on consumer AI services. This shadow AI usage creates uncontrolled FERPA exposure that administrators may not detect.
Cases involving third-party data sharing rose 34% in 2024, driven in part by the rapid expansion of educational technology. Much of this increase likely stems from well-intentioned staff using tools without understanding the compliance implications.
Where Student Data Goes When You Use AI
Understanding data flow is essential for compliance assessment.
Consumer AI Services (ChatGPT Free/Plus, Claude Free/Pro, Gemini Free)
When educators use consumer AI services with student data:
- Student information leaves the school's controlled environment
- Data travels to the AI provider's cloud infrastructure
- Information may be stored for varying retention periods
- Data may be used for model training (depending on settings and terms)
- Human reviewers may access conversations for quality control
- Data may flow to subprocessors for various services
Consumer services typically lack the contractual agreements FERPA requires for school official status.
Enterprise AI Services (ChatGPT Enterprise, Claude Enterprise, Microsoft 365 Copilot)
Enterprise tiers offer improved protections:
- Data Processing Agreements define how data is handled
- Training opt-outs or no-training commitments may apply
- Data residency options may be available
- Audit logging provides visibility into usage
- Access controls limit who can process data
However, enterprise agreements must still be evaluated against FERPA's specific requirements. Having an enterprise license doesn't automatically confer school official status on the vendor.
API and Cloud-Deployed AI
API access through cloud providers like AWS, Google Cloud, or Azure may offer the most control:
- Data remains within your cloud environment
- Processing occurs under your existing cloud agreements
- Retention can be configured to your requirements
- No training on your data (in most configurations)
This approach requires technical implementation but offers the strongest compliance posture.
Building a FERPA-Compliant AI Workflow
The most reliable approach separates AI capability from student data exposure.
The Architecture
Student records in SIS/LMS
↓
Automated detection of FERPA-protected PII
↓
Redaction layer replaces identifiers with placeholders
↓
Sanitized content flows to AI processing
↓
AI generates response using placeholders
↓
Output reviewed by authorized personnel
↓
Final documents reconstituted in school systems
With this architecture, student PII never reaches the AI system. The information processed by the AI isn't education records because identifying information has been removed. FERPA's disclosure restrictions don't apply to properly de-identified data.
What Gets Redacted
Before any student data reaches AI processing, remove:
Direct identifiers:
- Student names
- Parent names
- Student ID numbers
- Social Security numbers
- Dates of birth (except year for aggregate analysis)
- Addresses and phone numbers
- Email addresses
- Photos and biometric data
Indirect identifiers:
- School names (if could identify student)
- Specific dates of incidents
- Unique course combinations
- Small group references (e.g., "the three students in advanced physics")
- Any detail that could identify a student in combination with other available information
Placeholder Conventions
Replace identifiers with consistent placeholders:
[STUDENT-1],[STUDENT-2]for student names[PARENT-1]for parent names[SCHOOL-1]for school names[DATE-1]for specific dates[GRADE-X]for grade levels[INCIDENT-1]for incident references
Maintain consistency throughout documents so the AI can track references while maintaining anonymity.
Implementation Checklist
Step 1: Inventory Current AI Usage
Survey faculty and staff to understand:
- What AI tools are being used?
- What types of student data are being processed?
- Are personal accounts or school-approved tools being used?
- What workflows involve student information?
Expect to discover shadow AI usage. The goal is to understand the current state before implementing controls.
Step 2: Classify Data by Sensitivity
Create tiers based on FERPA risk:
Highest risk (mandatory redaction):
- Special education records
- Disciplinary records
- Health information
- Counseling notes
- Financial information
High risk (strong redaction recommended):
- Individual student academic records
- Attendance records
- Communications about specific students
Lower risk (aggregate data, properly de-identified):
- Curriculum materials without student specifics
- General educational content
- Administrative documents without PII
Step 3: Evaluate AI Vendor Agreements
For any AI tool used with student data, verify:
- Does the vendor agreement meet FERPA's school official requirements?
- Is the school maintaining direct control over data use?
- Is data use limited to authorized purposes?
- Are appropriate security safeguards in place?
- What are the data retention and deletion policies?
- Does the vendor use data for training?
- Who are the subprocessors?
Document this evaluation for compliance records.
Step 4: Implement Technical Controls
Deploy automated redaction that:
- Scans all content before AI processing
- Detects student names through NER and pattern matching
- Identifies student IDs, SSNs, and other structured identifiers
- Flags potential indirect identifiers for review
- Replaces detected PII with consistent placeholders
- Logs all redaction actions for audit purposes
Step 5: Establish Governance Policies
Create and enforce policies covering:
- Which AI tools are approved for which use cases
- What data categories require redaction before AI processing
- Who is authorized to use AI with student information
- How AI-assisted work must be documented
- Incident response procedures for accidental disclosure
Step 6: Train Faculty and Staff
Ensure all personnel understand:
- Why FERPA applies to AI usage
- What constitutes student PII
- How to use approved redaction workflows
- When to escalate questions to compliance officers
- Consequences of unauthorized AI usage
A recent survey found that while educators are rapidly adopting AI, many lack formal training on the privacy implications.
Audit Trail Requirements
FERPA requires schools to maintain records of disclosures of PII from education records. When AI is involved in processing student information, audit requirements expand:
What to Document
AI usage records:
- Who used AI tools
- What category of student data was involved
- When processing occurred
- What redaction was applied
- What outputs were generated
Vendor compliance:
- Vendor agreements and their FERPA provisions
- Security assessment results
- Subprocessor documentation
- Incident reports and responses
Training records:
- Who received FERPA/AI training
- When training occurred
- What topics were covered
- Assessment results
Retention Periods
Maintain disclosure records for as long as the education records themselves are maintained. For AI usage logs, implement retention policies that balance compliance needs with storage constraints. Consider:
- Detailed logs for recent activity (1-2 years)
- Summary records for longer periods
- Preservation of incident-related logs indefinitely
The Compliance Trajectory
FERPA enforcement is evolving. The Department of Education's Family Policy Compliance Office has adopted a more aggressive approach, treating data protection as a priority for schools of every size. Cases involving third-party sharing rose 34% in 2024.
In April 2025, the White House issued requirements for schools to implement AI resources and policies within 180 days. A growing body of commentary notes that FERPA lacks clear cybersecurity requirements, and experts urge lawmakers to modernize the law with vendor security obligations.
While FERPA's ultimate penalty (withdrawal of federal funding) has never been imposed, the reputational damage from student data breaches is severe. The PowerSchool incident affected thousands of school districts and dominated education news for months. Schools that are proactive about AI compliance avoid becoming the next headline.
Practical Use Cases
Here's how FERPA-compliant AI workflows look in practice:
Progress Report Drafting
Risky workflow: Paste student academic records into ChatGPT to generate progress report comments.
Compliant workflow:
- Export student performance data from SIS
- Run through redaction to replace names and identifiers with placeholders
- Submit to AI: "Write progress report comments for [STUDENT-1] based on these academic indicators"
- Review AI output with placeholders
- Import into SIS where your system maps placeholders to actual students
IEP Development Support
Risky workflow: Upload student evaluation data to AI for IEP goal suggestions.
Compliant workflow:
- Redact all student and family identifiers from evaluation data
- Replace disability classifications with codes or placeholders
- Submit: "Suggest measurable IEP goals for [STUDENT-1] based on these assessment results for [DISABILITY-TYPE]"
- Review and customize suggestions within your secure IEP management system
Behavioral Incident Analysis
Risky workflow: Share incident reports with AI to identify patterns.
Compliant workflow:
- De-identify all incident reports, removing student names, dates, and identifying details
- Aggregate data where possible to prevent re-identification
- Submit anonymized patterns to AI for analysis
- Apply insights at the policy level without exposing individual student records
The Bottom Line
FERPA applies fully to AI usage with student education records. Consumer AI services generally cannot meet FERPA's requirements for school official status. Even enterprise AI tools require careful evaluation of vendor agreements and data handling practices.
The path to compliant AI usage in education:
- Assume shadow AI is happening in your schools
- Inventory and classify AI usage by FERPA risk
- Evaluate vendor agreements against FERPA requirements
- Implement automated redaction before AI processing
- Establish governance policies and enforce them
- Train all personnel on compliant workflows
- Maintain audit trails documenting AI usage
The 60 million students affected by the PowerSchool breach trusted that schools and vendors would protect their information. When educators adopt AI tools without proper controls, they're adding new vectors for the exposure that students and families trust schools to prevent.
AI tools can genuinely help educators. Lesson planning, progress reporting, administrative tasks: these use cases offer real productivity gains. But those benefits must be balanced against FERPA's mandate to protect student privacy.
Get the compliance infrastructure right, and AI becomes a powerful tool that enhances education while protecting students. Get it wrong, and you're adding to the breach statistics that already make education one of the most targeted sectors in cybersecurity.
PaperVeil lets you redact sensitive information from documents before they touch any AI system. Detect and remove student identifiers, family information, and education records automatically. Generate the audit trails that FERPA compliance requires. The redaction layer that makes AI document processing actually safe for education.