ChinaMeasures for AI Science and Technology Ethics Review and Services (Trial)
In force since .
An AI risk obligations rule binding private bodies.
- How the hook was established
- express
- What makes it apply
- Operator establishment
- Obligation class
- DPIA, Governance
- Audit expectation
- periodic
- Who audits it
- Internal independent
- Where the report goes
- Filed with regulator
What it requires
- Applies to universities, research institutions, medical and health institutions, enterprises and other units that carry out AI science research and technology development in China that may bring science and technology ethics risks in areas such as human dignity, public order, and life and health (Arts. 2 and 9).
- Set up an AI science and technology ethics committee with the staff, premises and funding it needs and measures that let it work independently; its members must include experts in AI technology, applications, ethics and law (Arts. 9 and 10).
- Before carrying out a covered AI activity, apply for ethics review to the unit's committee, or to a service center where the unit has no committee or its committee cannot do the review, submitting the activity plan, the ethics risk assessment with monitoring, prevention and emergency plans, and a written undertaking (Art. 12).
- Identify changes in ethics risk in time and report them to the committee or service center, which follows up on approved activities generally at intervals of no more than 12 months (Art. 17).
- For an activity on the expert re-review list (human-machine integration systems with a strong influence on behavior, mental state and health; algorithm models, applications and systems able to mobilize public opinion; highly autonomous automated decision systems in settings with safety or health risks), apply for expert re-review after the preliminary review; follow-up review of such an activity is generally at intervals of no more than 6 months (Arts. 21, 22 and 25 and the Annex).
- Register the committee and the activities on the re-review list on the national science and technology ethics management information registration platform and submit the previous year's committee work report and the report on the implementation of listed activities (Art. 30).
- Provide channels for reporting violations of AI science and technology ethics and handle reports under the rules (Art. 31).
These lines are LexLint's reading of the law's text.
Who enforces it
Enforcement body
Ministry of Industry and Information Technology, with the Ministry of Science and Technology and other departments within their duties
What record it asks for
What the log must hold
Data lineage
Logging duty
The Measures name a log as a point the committee reviews and set no content list, retention period or access rule.
- Kind
- Implicit
- As of
- Provision
- Art. 15(5): ethics review looks at whether log management and other measures clearly record sufficient information on data, algorithms, models and systems at every stage
- Trigger
- all_systems
What this law does
The AI Science and Technology Ethics Review and Services Measures (Trial), issued under document number 工信部联科〔2026〕75号 by the Ministry of Industry and Information Technology and nine other bodies, apply to artificial intelligence research and technology development carried out in China that may bring science and technology ethics risks and challenges in areas such as human dignity, public order, and life and health.
Universities, research institutions, medical and health institutions and enterprises that carry out AI science and technology activities are responsible for ethics review management in their own unit and must set up an AI science and technology ethics committee. The person in charge of a covered AI activity must apply to the unit's committee, or to a service center where the unit has no committee or its committee cannot do the review.
A committee or service center must decide within 30 days of accepting an application whether to approve, send it back for revision and re-review, or refuse approval. An AI activity on the expert re-review list must, after the preliminary review by the committee or service center, be put forward by its unit for expert re-review.
The re-review list covers research and development of human-machine integration systems with a strong influence on human behavior, mental state and health, of algorithm models, applications and systems able to mobilize public opinion and society, and of highly autonomous automated decision systems in settings with safety or health risks.
Units must register their committee and the activities on the re-review list on the national science and technology ethics management information registration platform and submit the previous year's committee work report. Violations are investigated and punished under the Cybersecurity Law, the Data Security Law, the Personal Information Protection Law, the Law on Progress of Science and Technology and other laws, regulations and rules. The Measures take effect from the date of issuance.
When LexLint raises it
When your app profile says your app trains models.
Back to the example · Lint your app
Page updated .