Latest
  1. Johns Hopkins Bayview tells 3,182 patients an employee put their health data into a public AI tool
  2. China’s AI compute access expected to double in 2027, with US-designed chips about 65 per cent
  3. UK commits £100m to national AI Biosecurity Lab
  4. Moonshot AI releases Kimi K3.1 Pro with 1-million-token context window
  5. Residents in Newport and Cardiff oppose hyperscale datacentres near homes
  6. Study: Grok 4.7 picked Republicans in 92% of forced midterm answers, GPT-6.1 Sol picked Democrats in 97%
  7. Attackers exploiting the PHP flaws hand the break-in to AI agents, Unit 42 says
  8. Elvis Presley’s rightsholders sue Anthropic over training on 274 songs

Latest / Frontier models and safety

ReportFrontierGovernmentChinaConfirmed

Only 3.6% of 857 Chinese model releases ever got a published safety result

See how this story connects

Source: SemiAnalysis

SemiAnalysis built a dataset of 857 model releases from nine Chinese developers between 2021 and 15 September 2026 and found that 31, or 3.6%, ever had a developer safety result published. Nine had one at or before launch, 16 appeared later (a median 42 days after release) and six could not be timed. Another 10 releases came with an evaluation claim but no figures and 3 are known only from press accounts, leaving 813 with no safety disclosure at all. Alibaba published a result for 7 of 238 releases, Tencent 1 of 133, ByteDance 2 of 120 and Baidu 1 of 49; the five start-ups disclosed at 6.3% against 2% for the four hyperscalers. About 93% of reasoning models had none. Three disclosures touched cyber-offence or biological risk, and none of those came from a hyperscaler. Claims that a model was safety-trained or evaluated were not counted.

Why it matters

Chinese labs publish open weights anyone can run, and this counts how rarely any safety result is published alongside them.

When
8 Oct 2026
ID
WTA-N-2026-0248
Where
China
Who
SemiAnalysis, Alibaba, Tencent, ByteDance, Baidu, DeepSeek, Moonshot AI, Zhipu AI, MiniMax, StepFun; Dylan Patel

Summary by wheresthe.ai, from SemiAnalysis. Checked against the source on 12 Oct 2026.

Read the original at SemiAnalysis

Source copy: download the page as saved on 18 Oct 2026 (PDF). Also held by the Internet Archive.

Next stop on this Storyline: What happened nextDeepSeek and Huawei narrow AI chip gap with Nvidia, SemiAnalysis says

Cite this story

WTA-N-2026-0248: “Only 3.6% of 857 Chinese model releases ever got a published safety result.” wheresthe.ai, 8 Oct 2026. https://wheresthe.ai/d/cn-semianalysis-safety-disclosure-study-2026-10/ (source: SemiAnalysis).

Spotted an error? Email [email protected].