---
title: "China's privacy audit rule is now in force: 10 million people's data triggers a biennial check"
date: 2026-11-15
category: Policy & Governance
site: NeuroAI
canonical: https://neuroai.site/a/na-policy-personal-info-compliance-audit
language: en
---

# China's privacy audit rule is now in force: 10 million people's data triggers a biennial check

> The Personal Information Protection Compliance Audit Measures took effect on 1 May 2025, requiring handlers of 10M+ people's data to be audited every two years.

Imagine a compliance officer at a major Chinese super-app opening an email that reshaped her annual plan: a new rule meant her company, processing the personal information of more than 10 million people, now had to be audited for privacy compliance at least once every two years — and the state could order an audit at any moment if something went wrong.

The rule is not a proposal. It is live. The Personal Information Protection Compliance Audit Measures (个人信息保护合规审计管理办法), issued as CAC Order No. 18 (国家互联网信息办公室令 第18号), took effect on 1 May 2025. For the first time, China turned the vague obligation to "protect personal information" into a scheduled, verifiable chore.

## How a rule moves from paper to duty

The Measures sit at the back end of a chain. The Personal Information Protection Law (个人信息保护法, PIPL) and the Network Data Security Management Regulations (网络数据安全管理条例) required handlers to conduct compliance audits; the Measures, adopted at a CAC executive meeting on 20 May 2024 and promulgated on 12 February 2025, fill in the how, the who, and the how often.

A compliance audit here means a review of whether a personal-information handler's processing activities actually obey the law. It is a "physical," not a slogan: documented procedures, identified gaps, and a report.

The Measures also attach an Audit Guidelines document (个人信息保护合规审计指引) that distills the key compliance points of the PIPL and the data-security regulations into a checklist. Handlers on either track must reference it — it is the practical yardstick an auditor measures against, turning abstract legal duties like "minimum necessary" and "purpose limitation" into itemized questions a firm can actually answer.

## Who must audit, and how often

The core duty splits into two tracks.

- Self-initiated audit: handlers should regularly review their own compliance. The hard trigger is scale — any handler processing the personal information of more than 10 million people must conduct a personal information protection compliance audit at least once every two years.

- Triggered audit: regulators (the CAC and other protecting departments) may order a handler to commission a professional firm when (1) the processing poses a serious risk to individuals' rights or shows a grave lack of security measures; (2) it may infringe the rights of many individuals; or (3) a security incident has leaked, altered, lost, or destroyed the personal information of more than 1 million people, or the sensitive personal information of more than 100,000 people.

The 10-million line is the headline. It captures the super-apps, the big e-commerce and ride-hailing platforms, and the major financial and health-data holders — the entities whose databases, if mishandled, affect the largest slice of the population.

## What the auditors themselves must do

The Measures pay unusual attention to the people doing the auditing, not just the audited. A professional institution must have real capacity — qualified staff, premises, facilities, and funding — and is encouraged to seek certification. Three constraints stand out:

- No sub-delegation: a firm hired to audit cannot pass the work to another institution.

- Independence cap: the same institution and its affiliates, and the same audit lead, may not audit the same object more than three times in a row (连续三次以上).

- Confidentiality: anything learned — personal information, trade secrets, restricted business information — stays locked.

The "three-times" rule is the clever bit. It stops the comfortable, years-long client relationship in which an auditor quietly stops seeing what it is paid to see. Rotation is built into the regulation.

## Why it matters beyond a checkbox

For companies, the practical effect is a recurring cost and a recurring risk. A triggered audit is paid for by the handler, must finish inside a regulator-set window, and ends in a report plus remediation. For users, it is the difference between a privacy promise and a privacy receipt: someone, independent, has to look.

Globally, this is not exotic. The U.S. Federal Trade Commission has long extracted third-party privacy audits through settlement agreements; the EU's GDPR builds in auditing logic. What China's Measures add is scale-based compulsion and a built-in rotation rule, attached to the world's largest concentration of consumer data.

## Honest limitations

This article describes the Measures as in force since 1 May 2025; subsequent guidance or local enforcement practice may have refined how terms like "serious risk" are applied, and I have not audited individual enforcement cases. My facts come from the CAC's official text and corroborating state-media reporting (Xinhua, People's Daily, China Youth Daily). The threshold figures (10 million; 1 million; 100,000) are taken verbatim from the order and should be read as legal lines, not approximations. Penalty amounts for violation are set elsewhere in the PIPL and the regulations, not spelled out in the Measures themselves, so I do not quote a fine figure here.

## What readers can do now

- If your product touches the personal data of more than 10 million people in China, put a biennial audit on the calendar now — do not wait for a regulator's letter.

- Map your breach thresholds: a leak of 1 million records or 100,000 sensitive records can trigger a state-ordered, handler-paid audit with a fixed deadline.

- Treat the audit as evidence, not paperwork — keep the report and your remediation visible, because the same institution cannot audit you more than three times running, and the next firm will start from your last file.

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-policy-personal-info-compliance-audit
Free to quote with attribution and a link to the original.
