--- type: "post" id: "on-thursday-research-firm-semianalysis-published-a-report-5ef491" slug: "on-thursday-research-firm-semianalysis-published-a-report-5ef491" title: "On Thursday, research firm SemiAnalysis published a report analyzing hundreds of model releases from nine leading Chinese labs over the past five years, finding that only a small percentage of them..." description: "SemiAnalysis’s review makes the release-level safety record the object of measurement." retrieval_nugget: "SemiAnalysis’s review makes the release-level safety record the object of measurement." published_at: "2026-10-12" updated_at: "2026-10-12" record_date: "2026-10-12" date_kind: "published_at" topics: [] source_urls: ["https://newsletter.semianalysis.com/p/beijing-will-not-pace-the-frontier"] schema_version: "newruntime-agent-readable-v0.2" stable_id: "post:on-thursday-research-firm-semianalysis-published-a-report-5ef491" status: "published" visuals: [] editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:d7d5346697084351d9711d8b95ef78b0d40770d1cb08b6785812befb59c96aa8","reviewed_at":"2026-10-12T12:02:45.709Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":2,"watch_condition_count":1,"related_record_count":0} analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Safety disclosure is only informative when an evaluation result can be matched to the specific model released.","observed_facts":[{"text":"SemiAnalysis says its dataset contains 857 counted releases from nine leading Chinese developers, from 2021 through 15 September 2026.","source_urls":["https://newsletter.semianalysis.com/p/beijing-will-not-pace-the-frontier"]},{"text":"The dataset covers ByteDance, Alibaba, Tencent, Baidu, DeepSeek, Moonshot, Zhipu Z.ai, MiniMax, and StepFun.","source_urls":["https://newsletter.semianalysis.com/p/beijing-will-not-pace-the-frontier"]},{"text":"The stated check matches model cards, release notes, and technical reports to results on harmful output, jailbreaks, toxicity, privacy, refusal, or dangerous capability.","source_urls":["https://newsletter.semianalysis.com/p/beijing-will-not-pace-the-frontier"]}],"mechanism":"The mechanism is a model-by-model evidence check rather than a count of broad safety commitments: a result must be a quantitative or substantive finding on harmful output, jailbreaks, toxicity, privacy, refusal, or dangerous capability, tied to the named model.","why_now":"The report examines a five-year release record across both major hyperscalers and startups, making disclosure practice a concrete research question rather than a debate about stated principles alone.","implications":["Safety reporting can be assessed as part of release evidence, not inferred from a lab’s general posture.","Matched findings make disclosure comparisons more meaningful, but do not by themselves reveal internal evaluations that were not published."],"evidence_boundary":"The supplied excerpt describes the dataset and test but not the final disclosure rate; it also covers nine Chinese developers, not US labs.","watch_conditions":["Whether the full report publishes the release-level findings and enough detail for others to reproduce or challenge the matching criteria."],"related_records":[],"new_branch_reason":"No prior New Runtime record for this report is supplied; this opens a distinct record on release-level safety-evaluation disclosure."} editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]} routes: {"html":"https://newruntime.com/posts/on-thursday-research-firm-semianalysis-published-a-report-5ef491/","markdown":"https://newruntime.com/posts/on-thursday-research-firm-semianalysis-published-a-report-5ef491.md","json":"https://newruntime.com/posts/on-thursday-research-firm-semianalysis-published-a-report-5ef491.json"} --- # On Thursday, research firm SemiAnalysis published a report analyzing hundreds of model releases from nine leading Chinese labs over the past five years, finding that only a small percentage of them... ## Retrieval answer SemiAnalysis’s review makes the release-level safety record the object of measurement. SemiAnalysis’s review makes the release-level safety record the object of measurement. Safety disclosure is only informative when an evaluation result can be matched to the specific model released. The firm says it assembled a dataset of 857 model releases from nine leading Chinese developers, covering releases from 2021 through 15 September 2026, and checked model cards, release notes, and technical reports. The mechanism is a model-by-model evidence check rather than a count of broad safety commitments: a result must be a quantitative or substantive finding on harmful output, jailbreaks, toxicity, privacy, refusal, or dangerous capability, tied to the named model. The report examines a five-year release record across both major hyperscalers and startups, making disclosure practice a concrete research question rather than a debate about stated principles alone. Safety reporting can be assessed as part of release evidence, not inferred from a lab’s general posture. The supplied excerpt describes the dataset and test but not the final disclosure rate; it also covers nine Chinese developers, not US labs. Whether the full report publishes the release-level findings and enough detail for others to reproduce or challenge the matching criteria. The report establishes a measurement approach; it does not, from the excerpt available here, establish that non-disclosure means no internal safety work occurred. Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json ## Continue in the graph Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`. - collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis. - collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph. - whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)