Overview & Analysis
The Measures for the Certification of Cross-Border Personal Information Transfers establish one of three compliance pathways under China's personal information export framework — alongside security assessment and standard contracts — allowing qualifying personal information processors to transfer personal data overseas by obtaining certification from an accredited professional certification body. The Measures apply specifically to non-CII operators transferring personal information at mid-range volume thresholds: more than 100,000 but fewer than 1,000,000 individuals' non-sensitive personal information, or fewer than 10,000 individuals' sensitive personal information, cumulatively in a calendar year. Important data is explicitly excluded from this pathway. Processors must complete notice obligations, obtain separate individual consent, and conduct a personal information protection impact assessment before applying for certification. Certificates are valid for three years and require renewal applications six months before expiry.
This regulation is highly important for AI development, particularly when AI projects involve cross-border data flows. AI projects often rely on global data sharing, especially for large-scale data exchanges in model training and inference processes. The certification route ensures that cross-border AI projects comply with China's personal information protection laws while reducing compliance risks compared to the full security assessment pathway. This approach helps companies overcome obstacles in cross-border data transfer and provides a legal framework for the operation of global AI platforms — which is particularly crucial for AI applications that handle personal and sensitive data across multiple jurisdictions. The Measures also clarify that processors cannot circumvent higher-threshold security assessment requirements by splitting data volumes to artificially fall within certification thresholds.
These Measures govern the certification pathway only. Understanding where it sits relative to the other two pathways is essential for route selection in AI projects:
| Pathway | When Required / Available | Key Trigger Thresholds |
|---|---|---|
| Security Assessment | CII operators; or processors reaching higher volume thresholds; or important data | ≥1M individuals' PI; or ≥100K sensitive PI; or any important data; or CII operators |
| Certification (this document) | Non-CII operators within mid-range thresholds; no important data | >100K but <1M individuals' PI; or <10K sensitive PI (cumulative in calendar year) |
| Standard Contract | Non-CII operators within lower thresholds | <100K individuals' PI; or <10K sensitive PI (cumulative in calendar year) |
In AI applications, scenarios involving cross-border data transfers are the most relevant, especially when AI projects rely on global data sharing for model training, inference, business optimization, and so on. Any AI project involving the cross-border flow of personal or sensitive information that meets the conditions set out in these Measures must transfer the data via the certification pathway.
1. Global AI Model Training and Inference
AI systems require large amounts of data for training and inference, making cross-border data transfer essential. If the data involves personal or sensitive information and falls within the certification thresholds, companies must obtain certification for data transfer and ensure that data processing complies with China's personal information protection standards. The certification pathway provides a structured but more flexible route compared to the security assessment for AI teams operating at mid-range data volumes.
2. Using Overseas Cloud Services and Cross-Border Technology Platforms
When AI projects use overseas cloud computing resources or technology platforms, personal or sensitive information involved must follow an approved cross-border transfer pathway. The certification route provides a clear compliance path for multinationals using global cloud platforms, ensuring secure and lawful data transfers while avoiding the heavier burden of full security assessment where thresholds permit.
3. Sharing Customer Data in Cross-Border Business Operations
Cross-border AI applications — particularly in multinational marketing, customer service, and data analytics scenarios — inevitably involve cross-border customer data transfers. For data volumes within the certification thresholds, this pathway allows organizations to ensure that personal information is protected during transfer through a professionally certified conformity assessment rather than a government-led security review.
4. Cross-Border HR Management and Employee Data
Multinational companies often rely on AI for human resources management, which involves transferring employee data across borders. Companies need to assess whether employee personal information (including sensitive categories such as health data, biometrics, or financial account information) falls within the certification threshold or triggers higher-level security assessment requirements, and design their HR data architecture accordingly.
5. International R&D Collaborations and AI Data Sharing
Cross-border collaborations and joint R&D projects involving AI technologies typically require sharing sensitive data and technologies. The certification route allows companies to ensure that data shared with international partners complies with security requirements, safeguarding compliance with China's data protection laws while maintaining the flexibility needed for international research partnerships — provided data volumes stay within the prescribed thresholds.
These Measures provide multinational companies with a simplified compliance pathway, transforming cross-border data flow from a complex legal barrier into an operational issue that can be managed flexibly through professional certification. For AI project managers, this policy clarifies the compliance requirements for cross-border data transfer, especially when handling personal and sensitive data.
Map Your Data Volumes to the Right Pathway Before Project Design
The three pathways — security assessment, certification, and standard contract — are triggered by cumulative annual data volumes and data type. Before selecting a pathway, map out all personal information flows that will cross the border in a calendar year: non-sensitive PI volume, sensitive PI volume, and whether any important data is involved. The certification pathway covers the mid-range band. Do not assume the same pathway applies to all transfers; different project streams may fall into different categories.
Complete Pre-Application Obligations Before Seeking Certification
Before applying for certification, processors must fulfill three prerequisites: provide notice to individuals, obtain separate individual consent for the specific transfer, and conduct a personal information protection impact assessment (PIPIA). The PIPIA must address the legality and necessity of the processing, data sensitivity and national security risks, the overseas recipient's obligations and security capabilities, risk of post-transfer data incidents, and the overseas jurisdiction's legal framework. Do not treat these as post-certification items — they must be done first.
Build a Data Flow Monitoring and Record-Keeping System
Establish monitoring and recording systems for all cross-border data transfers — including the destination, data type, volume, purpose, and security measures for each transfer. Certification does not eliminate ongoing obligations: if the certified processor's outbound activities become inconsistent with the certification scope, the certification body must suspend and ultimately revoke the certificate. Real-time awareness of actual transfer volumes against declared certification scope is essential to avoid triggering suspension.
Design Flexible Compliance Pathways Based on Data Type and Sensitivity
Do not treat all cross-border data flows the same. Low-risk data that does not involve personal or sensitive information may not require any of the three pathways. Standard contract covers the lower band. Certification covers the mid-range band. Security assessment covers CII operators, high volumes, or important data. Companies should design appropriate compliance paths based on data nature, scale, and sensitivity — and review these classifications whenever AI project scope expands.
Strengthen Compliance Review with Overseas Recipients and Vendors
The PIPIA specifically requires assessment of the overseas recipient's obligations, management and technical security measures, and capabilities to protect the transferred data. For multinational companies, this means vendor due diligence for cloud services, technology platforms, external data processors, and model providers must address their ability to meet Chinese personal information protection standards. Include binding compliance clauses in all cross-border data transfer agreements and ensure recipients understand and can support China certification requirements.
These Measures offer a streamlined compliance framework for mid-range cross-border personal information flows. By completing the required pre-application steps, selecting the right certification body, conducting rigorous impact assessments, and maintaining ongoing monitoring, multinational companies can effectively reduce legal risks and build sustainable cross-border data pipelines for AI projects in China — while preserving the flexibility needed for global operations.
Complete Regulatory Text
Article Index
- Articles 1–4 — Purpose, Scope, Definitions & Standards
- Articles 5–7 — Eligibility Conditions, Pre-Application Obligations & Application Procedure
- Articles 8–11 — Certification Process, Certificate Management & Reporting
- Articles 12–13 — Certification Body Filing & Supervision
- Articles 14–19 — Confidentiality, Complaints, Supervision Interviews, Liability & Effective Date
(1) It is not a critical information infrastructure operator;
(2) Since January 1 of the current year, it has cumulatively provided: more than 100,000 but fewer than 1,000,000 individuals' personal information (excluding sensitive personal information) overseas; or fewer than 10,000 individuals' sensitive personal information overseas.
The personal information provided overseas as referred to in the preceding paragraph does not include important data.
Where laws, administrative regulations, or provisions of the national cyberspace administration provide otherwise, such provisions shall prevail.
Personal information processors shall not adopt methods such as splitting quantities to provide personal information overseas through certification where they are legally required to undergo a security assessment for outbound data transfers.
(1) The legality, legitimacy, and necessity of the purpose, scope, and method of personal information processing by the personal information processor and the overseas recipient;
(2) The scale, scope, type, and sensitivity of the personal information to be transferred overseas, and the risks such transfer may pose to national security, public interests, and personal information rights and interests;
(3) Whether the obligations undertaken by the overseas recipient, as well as its management and technical measures and capabilities, can ensure the security of the personal information transferred overseas;
(4) The risks of tampering, destruction, leakage, loss, or illegal use of personal information after it is transferred overseas, and whether channels for safeguarding personal information rights are smooth;
(5) The impact of the personal information protection policies and regulations of the country or region where the overseas recipient is located on the security of outbound personal information and on personal information rights and interests;
(6) Other matters that may affect the security of outbound personal information.
Where a personal information processor located outside the territory of the People's Republic of China applies for certification, the application shall be assisted by its dedicated institution established within China or its designated representative.
The validity period of a certification certificate shall be three years. Where continued use is required upon expiration, the personal information processor shall apply for certification six months before the expiration of the validity period.
The state market supervision and administration department and the national cyberspace administration shall establish a mechanism for sharing certification information.
Where the national cyberspace administration or relevant departments discover such circumstances in the course of personal information protection supervision, the certification body shall cooperate in suspending use until revocation.
The circumstances described in the preceding two paragraphs shall be publicly disclosed through the National Certification and Accreditation Information Public Service Platform.
(1) Certification qualifications obtained in the field of personal information protection;
(2) Professional work experience in data security and personal information protection over the past three years;
(3) Security background check materials for certification personnel;
(4) Implementation rules and work plans for personal information protection certification;
(5) Mechanisms for preventing personal information security risks;
(6) Ongoing supervision mechanisms to ensure that certified personal information processors' outbound activities comply with certification standards;
(7) Complaint handling and dispute resolution mechanisms;
(8) Other required materials.
The certification body shall be responsible for the authenticity of the filed materials.
After receiving the filing materials, the national cyberspace administration shall, together with the national data administration, review the materials. Where materials are complete, filing shall be completed within 30 working days and publicly disclosed; where incomplete, filing shall not be accepted, and the certification body shall be notified within 30 working days with reasons provided.
个人信息出境认证办法
（2025年10月14日公布，自2026年1月1日起施行，第20号令）
来源：中国网信网
条文索引
（一）非关键信息基础设施运营者；
（二）自当年1月1日起累计向境外提供10万人以上、不满100万人个人信息（不含敏感个人信息）或者不满1万人敏感个人信息。
前款所称向境外提供的个人信息，不包括重要数据。
法律、行政法规或者国家网信部门另有规定的，从其规定。
个人信息处理者不得采取数量拆分等手段，将依法应当通过出境安全评估的个人信息通过个人信息出境认证的方式向境外提供。
（一）个人信息处理者和境外接收方处理个人信息的目的、范围、方式等的合法性、正当性、必要性；
（二）出境个人信息的规模、范围、种类、敏感程度，个人信息出境可能对国家安全、公共利益、个人信息权益带来的风险；
（三）境外接收方承诺承担的义务，以及履行义务的管理和技术措施、能力等能否保障出境个人信息的安全；
（四）个人信息出境后遭到篡改、破坏、泄露、丢失、非法利用等的风险，个人信息权益维护的渠道是否通畅等；
（五）境外接收方所在国家或者地区的个人信息保护政策和法规对出境个人信息安全和个人信息权益的影响；
（六）其他可能影响个人信息出境安全的事项。
中华人民共和国境外的个人信息处理者申请个人信息出境认证的，应当由其在境内设立的专门机构或者指定代表协助进行申请。
认证证书的有效期为3年。证书到期需继续使用的，个人信息处理者应当在有效期届满前6个月提出认证申请。
国家市场监督管理部门与国家网信部门建立认证信息共享机制。
国家网信部门和有关部门在个人信息保护监督管理工作中发现获证个人信息处理者存在前款情形的，专业认证机构应当配合暂停其使用直至撤销相关认证证书。
前两款规定的情形，应当通过全国认证认可信息公共服务平台予以公布。
（一）取得的个人信息保护领域的认证资质情况；
（二）近3年从事数据安全、个人信息保护领域专业工作情况；
（三）专业认证机构人员安全背景审查材料；
（四）个人信息保护认证实施细则及工作计划；
（五）个人信息安全风险防范机制；
（六）对获证个人信息处理者进行的个人信息出境活动符合认证标准情况的持续监督机制；
（七）投诉受理和争议解决机制；
（八）其他需要提交的材料。
专业认证机构应当对所备案材料的真实性负责。
国家网信部门收到专业认证机构提交的备案材料后，会同国家数据管理部门对备案材料进行审核。材料齐全的，应当在30个工作日内予以备案并进行公示；材料不齐全的，不予备案，应当在30个工作日内通知专业认证机构并说明理由。
