Safety is an evaluation product: benchmarks, risk-management systems, and commitment frameworks that industry can adopt and be measured against.
Evaluation InstitutionChina
China Academy of Information and Communications Technology
中国信息通信研究院
The MIIT institute that built much of China's AI evaluation machinery: the AIIA alliance, the AI Safety Benchmark, and the Fangsheng benchmark system all run through it.
Importance
Assessed per aspect, band 0 to 3. There is no composite score.
Regulatory authoritynot assessed
Standards influence
1 · emerging
Its risk-governance reports and alliance frameworks circulate as reference documents for industry practice, short of formal standard status.
Technical capacity
3 · central
It operates the Fangsheng benchmark system with a dedicated safety dimension and reports expanding the AI Safety Benchmark past one million test data points across more than thirty safety categories.
Convening power
3 · central
It jointly founded AIIA, serves as its secretary-general unit, and used the alliance to launch the Safety Governance Committee, a risk-management system, and industry safety commitments.
International interfacenot assessed
What safety means here
Leadership
No leadership roles are recorded with a verifiable source.
History
- 2018-09-18
Records its role as joint initiator and secretary-general unit of AIIA.
- 2024-07-11
Fangsheng large-model benchmark system report published, with safety as a dedicated dimension.
- 2024-12-25
AI Risk Governance Report 2024 published, documenting the AIIA Safety Governance Committee and the AI Safety Benchmark.
- 2026-01-09
AI Safety Governance Research Report 2025 published, describing the expanded benchmark and the China AI Safety Commitment Framework.
Connections
Its recorded edges show the double role: affiliated with MIIT upward, founder of AIIA downward, and a member of the CnAISDA research network. analyst judgment
Affiliated With
- Ministry of Industry and Information Technology CAICT's official about page states that it is a research public institution directly under MIIT.
Establishes
- China AI Industry Development Alliance CAICT stated on its official site that, acting as the secretary-general unit, it jointly initiated the establishment of AIIA. This is recorded as an establishes relation with medium confidence because the source is explicit about joint, not sole, establishment.
Member Of
- China AI Development and Safety Network Tsinghua's official coverage lists CAICT among the member units of CnAISDA.
Selected outputs
Engagement fit
Assessmentanalyst judgment
- Plausible for
- Working-level exchange that spans regulation, evaluation, standards, and industry practice in one conversation.
- Constrained by
- Influential but not the final rulemaking authority; some outputs are interpretive or alliance-driven rather than binding.
Open questions
- Leadership could not be recorded: references naming CAICT's president and vice president did not resolve to verifiable sources.
Source appendix (6 records)
- caict-fangsheng-benchmark-report-2024
Research Report on Large-Model Benchmarking System 2024 · China Academy of Information and Communications Technology · 2024-07-11open
- caict-ai-safety-governance-report-2025
AI Safety Governance Research Report 2025 · China Academy of Information and Communications Technology · 2026-01-09open
- caict-aiia-founding-news-2018
CAICT and Gartner Release the 2018 World AI Industry Development Blue Book · China Academy of Information and Communications Technology · 2018-09-18open
- caict-ai-risk-governance-report-2024
AI Risk Governance Report 2024 · China Academy of Information and Communications Technology · 2024-12-25open
- report-china-standards-evaluation-2026-06-05
China Standards and Evaluation Research Brief · Asia AI Safety Atlas research brief
- caict-about-us
CAICT About Us · China Academy of Information and Communications Technology · 2018-04-28open
Last verified 2026-07-07