AI & Algorithm Governance

China's AI policy stays promotion-first, and the promotion layer now sits simultaneously at the cabinet table and the summit podium. Nothing in this batch tightens algorithm, model or content rules; the control layer remains quiet. What is new is where the build-out has reached: the upstream data-input layer is now being institutionalised through named national pilots, extending the same planning logic from compute down to data labelling ¹.

The clearest signal remains the Sept 11 State Council executive meeting chaired by Li Qiang, which made the computing-power network (算力网) a standalone item and called it "the foundational support for AI development." It ordered "reasonable layout, orderly and regulated" build-out, a multi-tier networked compute system, and the operative pairing of compute-power coordination and compute-network integration (算电协同、算网融合), backed by resource monitoring and dispatch, backbone optical-fibre construction, and accelerated green-power direct connection and source-grid-load-storage projects ². This is siting, power and scheduling policy — industrial planning, not censorship.

Provincial evidence shows the template running. Guizhou is the only first-batch regional node whose consortium includes a grid company (Guizhou Power Grid, alongside the three telecom branches), creating a "three-network hub" that knits the national compute network, the new-generation communications network and the power grid together ³. Heilongjiang, another first-batch node, anchors on China Mobile's Harbin 10,000-card cluster ⁴ and reports 430 provincial digital workshops and smart factories by end-2025 — up 212 since 2022, a 25.4% annual rise — with the share of enterprises digitising key business processes climbing from 34.6% to 51.8% ⁵.

The new piece is data annotation. The National Data Administration's second batch of national data-annotation innovation pioneer cities named Shenyang (industrial manufacturing) and Dalian (medical health), making Liaoning a "two cities, two domains" pilot province. Liaoning's 2026–2028 action plan targets high-quality datasets; the province reports 27,000 TB of annotated data, 334 high-quality datasets, more than 120 firms, over 20,000 workers and a 6.2-billion-yuan industry. Dalian's data-annotation park, opened last year, has drawn 19 firms including 金慧科技 (Jinhui Technology), 医通康合 (Yitong Kanghe) and 东软云科技 (Neusoft Cloud), backed by a 3,000-firm software cluster with 200,000-plus software professionals and nine top-tier hospitals generating 53 million annual patient visits ¹. Data labelling is being treated as AI's "cornerstone" and staffed like a manufacturing sector.

Internationally, Beijing has moved from theme-setting to named deliverables. The APEC energy ministers' meeting placed "AI and energy mutual empowerment" among three priority directions under China's chairmanship ⁶; Xi's BRICS package adds an AI open-source zone, a digital industry cloud platform and a youth sci-tech exchange plan aimed at helping Global South countries "benefit equally" ⁷. That is the counter-narrative to Washington's "small circle" supply-chain push, reinforced by Assistant Foreign Minister Liu Bin's capacity-building seminar for more than 30 countries at Peking University ⁸.

Diffusion keeps widening beneath the diplomacy. MIIT's software plan still targets high-standard coding tools reaching more than 20,000 major enterprises and 100 benchmark agent-based applications by 2028 ⁹. Tokens are becoming ordinary: Unisound's "token export" served Ethiopian cross-border e-commerce with over 10 billion tokens in week one ¹⁰, while a dumpling shop near Beijing University of Posts and Telecommunications hands out 10-yuan compute cards ¹¹. Talent is wired in through Macao's Huawei-backed HCIA-AI training ¹², a university AI-and-brain-science symposium of about 500 participants ¹³, and Liaoning's vocational competition adding AI, low-altitude economy and industrial-internet tracks across 88 events with 2,100-plus contestants ¹⁴.

The read for outside observers: China is institutionalising AI's full stack — compute, power, data and skills — as planned infrastructure, and turning multilateral chairmanships into delivery vehicles, while leaving domestic model rules untouched. Three things to watch: whether the State Council's compute-network instruction hardens into a named plan, budget line or agency; whether the BRICS AI open-source zone gets a launch date as China takes the BRICS chair; and whether the data-annotation pilot model spreads nationally as a named program.