DeepSeek · Alibaba · ByteDance
Only 3.6% of Chinese AI model releases came with a published safety test result
Only 31 of 857 model releases, or 3.6%, from nine leading Chinese AI developers were ever accompanied by a published, model-specific safety evaluation result, according to a report released October 8 by California-based research firm SemiAnalysis. Just 9 releases (1.1%) had results available at or before launch. The count covers 2021 through September 15, 2026, and the companies include Alibaba, ByteDance, Tencent, Baidu and DeepSeek. Only concrete findings on harmful output, jailbreaks, privacy or dangerous capabilities counted, not general claims that a model was “safety-trained.” The report measures public disclosure only, so a missing result does not mean a model went untested. Every size and snapshot counts as a release, and there is no comparable figure for US developers.
Sources
- SemiAnalysis, “Beijing Will Not Pace the Frontier: China’s Speed-First AI Safety Regime”, 31 of 857 releases (3.6%) with a published safety result, (newsletter.semianalysis.com)
- Reuters (Devdiscourse üzerinden), “China AI developers publish safety tests for just 3.6% of model releases, report finds”, Method and caveats, (devdiscourse.com)
About this story
This story was posted on Instagram by @jarrus.tech on Oct. 13, 2026.
Spotted an error in this story? [email protected] · Instagram
Short link: thejarrus.com/en/china-safety-tests
This story in Turkish: Çinli şirketlerin model yayınlarının yalnızca %3,6’sı için yayımlanmış bir güvenlik testi sonucu var
On the same topic
Chinese AI infrastructure company Infinigence AI raises about $344 million and prepares for a Hong Kong IPO
Infinigence AI closed a 2.3 billion yuan Series C round, and Bloomberg reported a confidential Hong Kong IPO filing.
The US is again letting AI companies share their models with Britain’s AI Security Institute before release
According to The Telegraph, the US dropped its request that AI companies hold new models back from Britain’s AI Security Institute.
CNBC: Small groups like METR, Apollo Research and Transluce are emerging as AI’s safety gatekeepers
AI labs are leaning on independent evaluators, CNBC reported, but there are still no standards for funding, access or reporting.