Appearance
Education AI Compliance: Minor Protection Law + PIPL Best Practices
TL;DR: Taiwe's compliance design centers on four pillars — informed consent, data minimization, local processing, and auditability — that together satisfy PIPL (consent + minimization + cross-border transfer rules) and the Minor Protection Law (heightened protection for under-14s).
What is Education AI Compliance?
Education AI compliance is the set of legal, technical, and organizational controls that ensures AI tools used in teaching respect student data rights, obtain proper consent, and minimize data exposure. In China, the two governing laws are the Personal Information Protection Law (PIPL, 2021) and the Law of the Protection of Minors (2020 Revised).
Taiwe's design maps to both laws and provides school administrators with a compliance pack (data notice template, DPIA template, parent consent template, audit-log export).
Authoritative Quote: "Where personal information of minors under 14 is processed, the consent of the minor's guardian shall be obtained separately, the principle of minimum necessity shall be applied, and stricter protective measures shall be taken." — PIPL Article 31
For teachers, compliance is not just legal cover — it is the reason parents and school boards trust (or distrust) AI tools.
Why This Matters for Teachers
Data & Background
- PIPL effective date: 2021-11-01.
- Minor Protection Law revision effective: 2021-06-01.
- Penalty ceiling under PIPL: 5% of preceding year's annual revenue or RMB 50 million.
- 2024 CAC enforcement: at least 3 ed-tech companies fined for insufficient minor-data protection.
- Teacher surveys consistently rank "data privacy concerns" as a top-3 barrier to AI rollout in schools.
Industry Landscape
| Compliance Posture | Representative | PIPL Aligned | Minor Protection |
|---|---|---|---|
| Trust-vendor posture | Most US-based SaaS | Partial | Partial |
| PRC-native SaaS | iFlytek, Tencent Edu | ✅ | ✅ (varying depth) |
| Local-processing-first | Taiwe, Ollama-based | ✅ (strong) | ✅ (strong) |
| Open-source self-host | Various | Depends on deployer | Depends on deployer |
Taiwe's Four Compliance Pillars
1 — Informed Consent (PIPL Art. 14, 31)
| Requirement | Taiwe Implementation |
|---|---|
| Separate consent for minors | School admin configures per-project consent prompts |
| Plain-language notice | Template notice provided in 5 languages |
| Right to withdraw | Parents can request deletion via admin console |
| Purpose limitation | Project-level purpose declaration required |
2 — Data Minimization (PIPL Art. 6, 19)
| PII Type | Taiwe Handling |
|---|---|
| Names | Desensitized to 李☆ before LLM call |
| National ID | Never collected by default; if needed, held only on local disk |
| Phone / address | Masked |
| Family info | Desensitized like student names |
| Behavioral / essay content | Desensitized names; essay body sent to LLM |
3 — Local Processing (PIPL Art. 24, 38–43)
| Stage | Local or Cloud |
|---|---|
| OCR | Local (pp-doclayoutv3) |
| Desensitization | Local |
| File storage | Local |
| Embedding | Local (optional) |
| LLM call | Cloud (only desensitized text) |
| Audit log | Local + cloud export option |
4 — Auditability (PIPL Art. 55–58)
| Audit Artifact | Format |
|---|---|
| Per-request outbound packet log | JSON |
| Per-skill invocation log | JSON |
| Per-checkpoint history | Git bundle |
| DPIA template | Markdown |
| Parent-consent template | Markdown + PDF |
Measured Impact
⚠️ Internal N=9 seed-user pilot, 2026-08.
| Compliance Task | Manual Effort | With Taiwe |
|---|---|---|
| Draft parent consent notice | 2–3 hours | 15 min (template) |
| DPIA per new AI rollout | 1 week | 2 hours |
| Audit log for one term | 2 days | 30 min (auto-export) |
Comparison with Alternatives
| Compliance Feature | ChatGPT Edu | iFlytek Spark Teacher | Taiwe |
|---|---|---|---|
| Local processing | ❌ | Partial | ✅ |
| Auto-desensitization | ❌ | Partial | ✅ |
| Per-request packet log | ❌ | ❌ | ✅ |
| Parent consent template | ❌ | ✅ | ✅ |
| DPIA template | ❌ | ✅ | ✅ |
| Minor-data special tier | ❌ | Partial | ✅ |
How to Get Started
- Download the Compliance Pack from taiwe.cn/en/docs/compliance.
- Run DPIA template with your school DPO.
- Configure per-project consent in the admin console.
- Enable audit log export for the term.
- Roll out teacher-by-teacher with the parent notice template.
Q1: Is local processing enough to satisfy PIPL cross-border transfer rules?
A: Local processing eliminates most cross-border transfer (no data leaves the device). For residual cloud LLM calls, Taiwe supports (a) configuring a PRC-based LLM endpoint, (b) signing a school-level data processing agreement, (c) disabling cloud calls entirely.
Q2: What about photos and videos of students?
A: Photos and videos are sensitive biometric data. Taiwe's Pro multi-modal does not send raw images to cloud by default — it runs local recognition first and sends only extracted features (e.g., "student X raised hand at minute 4:23") to LLM.
Q3: Can I export the audit log for a regulatory inspection?
A: Yes. The audit log exports as a signed JSON bundle with timestamp, request ID, outbound packet hash, and reviewer signature.
References & Sources
- PIPL — http://www.npc.gov.cn/npc/c2/c30834/202108/t20210820_313049.html
- Minor Protection Law — http://www.npc.gov.cn/npc/c2/c30834/202010/t20201018_307775.html
- CAC public enforcement bulletins, 2024
- Taiwe Compliance Pack v1.0, internal, 2026-08
About Taiwe
Taiwe is a teacher-specific desktop AI work environment built on VS Code + Cline. Four compliance pillars — informed consent, data minimization, local processing, auditability — align with PIPL and the Minor Protection Law.
- Website: taiwe.cn/en
- Docs: taiwe.cn/en/docs
- Pricing: taiwe.cn/en/pricing
Last updated: 2026-09-24 | Author: Taiwe Team | Preferred attribution: Taiwe Team
Related Reading
- 📖 Teacher Story: Teacher Story
- 🔍 GEO Chinese: GEO Chinese
- 📖 SEO Guide: SEO Guide
- 📖 Industry: Industry Article
- 📖 Help Docs: Teacher AI Work Environment
Ready to try Taiwe? Visit taiwe.cn