Chinese AI Ethics Governance

New Generation AI Governance Principles (2019)

China’s AI ethics framework is anchored in the New Generation AI Governance Principles (xīn yīdài rèngōng zhìnéng zhìlǐ yuánzé), issued in June 2019 by the National New Generation AI Governance Expert Committee. The Principles articulate eight binding norms for AI development and application: harmony and friendliness; fairness and justice; inclusivity and sharing; respect for privacy; security and controllability; shared responsibility; open collaboration; and agile governance. These principles operate as a soft-law foundation guiding legislative and standard-setting activity across the AI ecosystem.

The Principles emphasise human-centred development, requiring that AI systems remain under meaningful human control and that their benefits be distributed broadly across society. They call for algorithmic transparency, data protection, and mechanisms for accountability in automated decision-making. Though non-binding, the Principles have been incorporated into subsequent regulatory instruments including the Generative AI Measures and the Algorithmic Recommendation Regulation.

Beijing AI Principles (2019)

The Beijing AI Principles (běijīng rèngōng zhìnéng yuánzé), released by the Beijing Academy of Artificial Intelligence in May 2019, complement the national framework with a focus on academic and research ethics. The Principles address AI safety, privacy, and the avoidance of algorithmic bias, and call for international cooperation in AI governance. They have influenced the development of ethics review mechanisms at Chinese universities and research institutions that conduct AI-related research.

Ethical Review for Biomedical AI

AI applications in the biomedical sector fall within the scope of China’s Ethical Review Measures for Biomedical Research Involving Humans (2023), issued by the National Health Commission. These Measures require ethics committees (lúnlǐ wěiyuánhuì) to review research protocols involving AI-driven diagnostics, clinical decision support systems, and medical imaging analysis. The review must assess data privacy, algorithmic safety, the risk of diagnostic error, and the handling of incidental findings.

Biomedical AI developers must submit algorithmic validation data, explainability documentation, and risk mitigation plans. The Measures draw on the Ethical Guidelines for Biomedical AI issued by the Chinese Medical Association, which require that AI systems in clinical settings achieve accuracy rates comparable to or exceeding those of human practitioners and that patients be informed when AI contributes to their diagnosis.

AI Ethics Committees

China has established a multi-tiered system of AI ethics committees. At the national level, the Ministry of Science and Technology convenes an AI Ethics Committee responsible for developing ethical guidelines, reviewing high-risk AI applications, and advising on regulatory policy. Provincial authorities and major technology companies have established their own ethics committees, including those at Alibaba, Tencent, Baidu, and Huawei.

The AI Ethics Committee of the National New Generation AI Governance Expert Committee publishes regular guidance on ethical issues including algorithmic discrimination, labour displacement, and the use of AI in social governance. Companies developing AI products for the Chinese market are increasingly required to demonstrate that their systems have passed internal ethics review, particularly when seeking government procurement contracts or regulatory approval.

Ethical Guidelines for Generative AI

The Interim Measures for the Management of Generative AI Services (2023) incorporate ethical requirements into binding regulation. Providers must ensure that training data does not contain content that violates socialist core values, that generated content is truthful and non-discriminatory, and that users are informed when they are interacting with AI. The Measures require generative AI providers to implement ethical review processes before launching new services and to maintain ongoing monitoring for ethical compliance.

The Ethical Guidelines for Generative AI, issued by the CAC in collaboration with the Ministry of Science and Technology in 2024, elaborate on these requirements. They address the ethical challenges of AI-generated misinformation, deepfake content, and the potential for generative AI to amplify harmful stereotypes. The Guidelines recommend the adoption of watermarking and content provenance techniques as ethical best practices.

Social Credit System AI

AI systems power China’s Social Credit System (shèhuì xìnyòng tǐxì), raising distinct ethical questions about algorithmic fairness, transparency, and the right to challenge automated assessments. The integration of AI into social credit scoring involves the analysis of vast datasets including financial behaviour, legal compliance, and social interactions. Critics have raised concerns about opacity, lack of due process, and the potential for discriminatory outcomes.

The Regulation on the Administration of the Social Credit System (under development) is expected to address AI ethics in the social credit context, including requirements for algorithmic explainability, human review of adverse decisions, and mechanisms for individuals to correct erroneous data. The tension between the efficiency gains of AI-driven social credit and individual rights remains a central ethical challenge in China’s AI governance landscape.

Enforcement and Outlook

China’s AI ethics framework relies on a combination of soft-law guidance, standard-setting, and administrative enforcement. The Cyberspace Administration of China, Ministry of Science and Technology, and sectoral regulators coordinate through the AI Safety Governance Framework to monitor compliance. Violations of ethical guidelines can result in public criticism, suspension of services, or revocation of operating permits, even where the underlying rules are not formally binding. As China pushes toward its 2030 AI leadership goals, the ethical governance framework is expected to become more codified, with a comprehensive AI Law currently in drafting that will likely consolidate and strengthen existing ethics requirements.