China TMT: Bi-monthly Update – July and August 2025 Issue

This newsletter summarises the latest developments in Technology, Media, and Telecommunications in China with a focus on the legislative, enforcement and industry developments in this area.

If you would like to subscribe for our newsletters and be notified of our events on China Technology, Media, and Telecommunications, please contact James Gong at james.gong@twobirds.com.

Key Highlights

In July and August 2025, relevant regulatory authorities advanced the standardized development of industries through multi-dimensional approaches including legislation, enforcement, and industry guidance in key areas such as artificial intelligence, intelligent connected vehicles, platform governance, and minors' protection, promoting technological and digital empowerment for industrial innovation, with particular emphasis on the integration of AI with existing industries:

• Artificial Intelligence: At the legislative level, the Ministry of Industry and Information Technology (“MIIT”) and the Cybersecurity Standardization Technical Committee (“TC260”) respectively issued measures on ethical requirements for artificial intelligence (“AI”) scientific and technological activities and technical documents on large models in the government affairs field to guide the further compliant development of AI. On the enforcement front, the Cyberspace Administration of China (“CAC”) notified 93 newly filed and 74 registered generative AI services; Beijing and Shanghai respectively added generative AI service registrations and filings; Guangdong announced its second-quarter enforcement notifications and imposed delisting penalties on non-compliant AI applications; the Beijing Internet Court ruled that AI “deepfake” of celebrity voices for product endorsements constituted infringement, clarifying the boundaries of voice rights and merchants' joint liabilities. At the industry level, central, local, and international entities issued documents requiring the expansion of artificial intelligence applications: the State Council issued opinions requiring the in-depth implementation of the "AI+" action; the Shanghai Municipal Government Office and the Economic and Information Technology Commission (“EITC”) respectively issued relevant documents to guide the development of embodied intelligence industries and accelerate the implementation of the “AI+” action; the World Artificial Intelligence Conference issued a plan to guide global governance actions for artificial intelligence. In addition, 18 enterprises disclosed the practical results of the Artificial Intelligence Safety Commitments to promote new progress in AI security through self-governance.
• Intelligent Connected Vehicles: The Ministry of Science and Technology (“MST”) issued guidelines to guide the research, development, and application of driving automation technology in an ethical manner. At the same time, the China Association of Automobile Manufacturers notified 13 enterprises of the testing results regarding five security requirements for automobile data processing.
• Network Platform Governance: At the legislative level, the National Development and Reform Commission and the State Administration for Market Regulation (“SAMR”) respectively issued rules and guidelines to standardize the determination, labelling, competition, and charging behaviours of internet platforms. For advertising, TC260 issued guidelines to standardize the display and triggering of shake-to-activate advertisements; the Hangzhou SAMR issued guidelines to standardize interactive advertisements and promotions. On the enforcement front, the CAC launched multiple actions to respectively address online "black mouths" involving enterprises, malicious marketing chaos in the short video field, the online environment for minors, "self-media" publishing false information, and the online environment during the Chengdu World Games; SAMR announced 7 typical cases in the live e-commerce field to guide live e-commerce operators to operate in compliance with the law. Local relevant institutions also actively took action: the Beijing CAC disclosed the algorithm rule principles of major local online platforms; the Guangdong CAC issued notifications on online reporting handling and enforcement; the Hainan CAC launched the "Clear and Bright Free Trade Port" special action; the Hangzhou SAMR issued 5 typical cases in the field of internet advertising and promotion. In addition, the Internet Society of China issued guidelines to strengthen compliance management for the protection of user rights in internet application services.
• Protection of Minors: TC260 plans to issue a national standard to standardize technical requirements for minors’ mode on mobile internet.

Follow the links below to view the official policy documents or public announcements.

Legislative Developments

1. National Development and Reform Commission, in collaboration with SAMR and CAC, planned to issue rules to standardize the determination, labelling, and competitive behaviours of internet platform prices (23 August)

The National Development and Reform Commission, in collaboration with multiple departments, solicited public opinions on the Internet Platform Price Behaviour Rules (Draft for Comments), aiming to improve the price supervision mechanism for the platform economy and standardize price behaviours. The rules apply to price behaviours such as the sale of goods or provision of services through information networks within the territory by platform operators and intra-platform operators, emphasizing principles of legality, compliance, fairness, integrity, and transparency. First, the rules safeguard operators' independent pricing rights, clarifying that platforms must not force (or disguisedly force) merchants to lower prices or set unreasonable restrictions such as "lowest price on the same platform." Second, the rules strengthen clear price labelling and information disclosure, requiring dynamic pricing, subsidies, promotions, etc., to be prominently displayed, and estimated prices to explain their composition and sources of differences. Third, the rules strictly prohibit "big data price discrimination," collusion, manipulation, and other behaviours, and require prominent reminders and convenient cancellations for automatic renewals/deductions. Fourth, the rules emphasize compliance and data security responsibilities: platforms should handle personal information in accordance with the law, properly conduct algorithm filings and cooperate with security assessments and inspections, and must not use data, algorithms, platform rules, etc., to set different prices or charging standards for the same goods or services under the same transaction conditions without consumers' knowledge.

2. MIIT planned to issue measures to specify ethical requirements for artificial intelligence scientific and technological activities (7 August)

The MIIT, in collaboration with multiple departments, solicited public opinions on the Measures for the Management and Service of Artificial Intelligence Technology Ethics (Trial), aiming to refine and implement national technology ethics requirements in the artificial intelligence field. The measures specify that they apply to AI research and technology development activities conducted within the territory that may pose ethical risks to life and health, public order, etc., proposing principles such as promoting human well-being, respecting life and intellectual property rights, fairness and justice, and reasonable risk control. At the same time, the measures establish dual supports of "standards + service system mechanisms," improving the AI ethics standard system, building service supplies such as risk monitoring and early warning, testing and evaluation, and strengthening support for small and medium-sized enterprises. At the implementation level, research institutions, enterprises, etc., are the main responsible entities for ethics management in their units, and organizations with conditions should establish ethics committees or entrust professional service centres for reviews. Reviews are divided into three categories: general, simplified, and emergency procedures, focusing on verifying fairness and justice, transparency and explainability, controllability and trustworthiness, and responsibility traceability. Activities with higher social risks, such as human-machine integration, public opinion mobilization, and automated decision-making, are included in the "review list" for re-examination by competent authorities, and connected with existing regulations on deep synthesis, algorithm recommendations, generative AI, etc. The measures also specify supervision mechanisms such as registration and reporting, issue handling, and violation handling.

3. MST issued guidelines to guide the research, development, and application of driving automation technology in an ethical manner (23 July)

The MST issued the Ethical Guidelines for the Research and Development of Driving Automation Technology, aiming to promote the standardized conduct of driving automation technology research and development. The guidelines adhere to principles of "people-oriented, safety first, fairness and justice, and informed assurance," requiring the establishment of full-lifecycle privacy protection mechanisms covering data collection, processing, storage, sharing, and deletion, following minimum necessity, informed consent, and de-identification. The guidelines propose requirements graded by technological evolution: In the advanced driver assistance stage, the guidelines emphasize reducing operational complexity and clearly prompting boundaries and responsibilities; in the restricted-stage autonomous driving, the guidelines point out that major changes to autonomous driving system content should be clearly recorded and accessible to regulatory authorities in a timely manner, strengthening the explainability of decision-making behaviours and incorporating user preferences into driving strategies; in the unrestricted-stage autonomous driving, the guidelines require the overall system safety level to be no lower than human driving, with decisions having stability and decision logic having high consistency, allowing autonomous driving systems to predict the action intentions of relevant entities based on real-time driving environments. In addition, the guidelines require enterprises to conduct risk communication and popular science, prohibiting exaggerated propaganda and misleading marketing.

4. SAMR issues guidelines to standardize charging behaviours of online trading platforms (31 July)

The SAMR issued the Compliance Guidelines for Charging Behaviours of Online Trading Platforms, aiming to standardize platform charges and safeguard the legitimate rights and interests of operators. The guidelines define "platforms," "intra-platform operators," and "platform charges," requiring platforms to follow relevant laws, adhere to fairness, legality, and good faith, set reasonable prices, provide concessions to small and medium-sized merchants, and appropriately reduce or exempt charges in emergencies. Second, the guidelines require the establishment of a charging compliance system, incorporating charges into organization, systems, risk control, and training, and encouraging the implementation of compliance reporting mechanisms. Furthermore, the guidelines require strengthening transparency and disclosure, such as continuously displaying charging items, rules, standards, and details on the homepage, with modifications requiring at least 7 days of public notice and solicitation of opinions, and retaining historical versions for 3 years. At the same time, the guidelines refine boundaries for marketing, promotions, and deposits, requiring reasonable billing, negotiated sharing of preferential subsidies, assessment of deposit necessity, closed operations, and timely refunds. The guidelines also specify unreasonable charging behaviours, prohibiting eight categories such as repeated charges, charging without sufficient services, forcing or disguisedly forcing purchases and promotions, price discrimination, and disguised charges via deposits. Finally, the guidelines establish an operator objection negotiation mechanism and require platforms to cooperate with supervision and industry organizations to strengthen self-discipline.

5. TC260 issues guidelines to standardize the display and triggering behaviours of shake-to-activate advertisements (22 July)

The TC260 issued the Cybersecurity Standard Practice Guide - Security Requirements for Shake-to-Activate Advertisement Triggering Behaviours, aiming to standardize the display and triggering behaviours of shake-to-activate advertisements and protect users' right to know and personal information security. The guide establishes basic principles of "transparency, autonomy, and personal information protection," requiring prominent prompts that "shaking" is needed to jump, providing clearly visible one-click closure, and setting a master switch within the app, after which related interactions must not continue to be displayed if closed by the user. To reduce accidental triggers, the guide proposes parameterized requirements for trigger sensitivity, clarifying that non-voluntary actions such as walking, riding, or handling devices should not be determined as valid triggers. At the same time, the guide requires third-party SDKs to open parameters for enabling or disabling shake-to-activate functions to apps, and parameters must not be adjusted without user consent.

6. TC260 planned to issue technical document to strengthen security in the application of large models in the government affairs field (28 July)

The TC260 solicited public opinions on the Security Specifications for Government Affairs Large Model Applications (Draft for Comments), aiming to provide enforceable security baselines and evaluation methods for introducing and using large models in government affairs scenarios, and to emphasize departmental primary responsibilities. The document focuses on the full chain of "selection-deployment-operation": In the selection phase, filed models should be used, open-source models require license and integrity verification, API calls need certificate validation and anti-shelling counterfeiting, and the use of retrieval-augmented generation (RAG) technology based on external knowledge bases is encouraged to improve accuracy; in the deployment phase, centralized construction and unified protection are required, with vulnerability testing and hardening for hardware/software and third-party tools, strict minimum privileges and interface frequency controls, and disabling high-risk operations; in the operation phase, generated/synthesized content identification, internal review of authoritative information, prominent prompts of model limitations, and retention of manual backups are required, with logs retained for no less than 6 months. Appendix A of the specifications simultaneously proposes "security guardrail" capabilities and provides testing points for selection, deployment, operation, and guardrail effectiveness before going online, forming a practical security baseline for government affairs large model applications.

7. TC260 planned to issue national standard to standardize technical requirements for minors' mode on mobile internet (29 July)

The TC260 solicited public opinions on Cybersecurity Technology - Technical Requirements for Minors Mode in Mobile Internet, aiming to clarify the construction and linkage requirements for "minors' mode" on terminals, apps, and distribution platforms, and to strengthen age-appropriate content supply and personal information protection. The technical requirements propose supporting one-click enabling/exiting of minors' mode and parental verification on the terminal side, with only basic functional, emergency for minors' personal health protection, and filed educational apps available by default in minors' mode; on the app side, services are not provided by default from 22:00 to 06:00 the next day, with daily total usage times graded by age at 1 hour/2 hours, continuous use reminders every 30 minutes, setting minors' exclusive zones and age-appropriate content pools, and improving restrictions on social, live streaming, browsers, search, e-commerce, consumption, and external links; distribution platforms should automatically detect the minors' mode status of terminals and correspondingly enable or exit minors' mode based on the terminal's status, while distribution platforms need to mark age-appropriate downloads and support parental verification/exemptions, achieve linkage for status, time, and app management through interfaces provided by mobile intelligent terminals, and follow personal information protection requirements such as minimum necessity and encryption of sensitive information.

8. Hangzhou SAMR issued guidelines to standardize internet interactive advertisements and promotions (1 July)

The Hangzhou SAMR issued the Guidelines for Risk Control and Legal Affairs in Internet Interactive Advertisements and Promotions (Trial), aiming to standardize online advertisements and promotions delivered through interactive methods, and to maintain fair competition and consumer rights. The guidelines specify: Interactive advertisements must be identifiable, with truthful and comprehensive content, prominently displaying activity rules and entity information, and must not mislead or induce with "today's fortune," "100% winning," etc., nor assist or tolerate false promotions. The guidelines stipulate that online lotteries must fully disclose lottery rules, prizes, probabilities, and redemption regulations, and should retain complete records of winning probability settings, winning results, etc. Finally, advertisements should ensure "one-click closure" and must not induce jumps with high-sensitivity "shake-to-activate," etc.; for those involving automatic renewals or deposit refunds, prominent prompts must be made for consumer selection, and convenient paths for cancellation and deposit refunds provided.

Enforcement Developments

9. CAC released progress on the “Clear and Bright· Optimizing the Business Network Environment” special action, addressing online “black mouths” involving enterprises and disclosing typical cases (2 July)

The CAC notified the progress of the “Clear and Bright· Optimizing the Business Network Environment—Addressing Online ‘Black Mouths’ Involving Enterprises” special action, exposing and handling a batch of accounts that fabricated rumours involving enterprises, extorted for profit, leaked trade secrets, and impersonated enterprise entities, emphasizing platform governance responsibilities, safeguarding enterprises' legitimate rights and interests, and maintaining online dissemination order. Typical violations include: fabricating product quality issues and defaming enterprises and entrepreneurs' images; using post deletion threats for “business cooperation”; hyping intelligent driving hotspots, fabricating accidents and judgments, and disparaging product quality; releasing enterprise trade secrets and spreading false information; impersonating enterprises or entrepreneurs for marketing. The aforementioned accounts have been closed or handled in accordance with laws and agreements, with relevant entities included in platform blacklists and infringing information cleaned up.

10. CAC issued announcement on filed generative artificial intelligence services, adding 93 filed AI services and 74 registered AI services (11 July)

The CAC issued the announcement on filed generative artificial intelligence services from April to June 2024. From April to June, 93 generative artificial intelligence services were newly filed with the national CAC; for generative artificial intelligence applications or functions that directly call filed model capabilities through API interfaces or other methods, registrations are conducted by local CACs, with 74 newly completed registrations in this phase. As of June 30, 2025, a cumulative total of 439 generative artificial intelligence services have completed filing, and 233 generative artificial intelligence applications or functions have completed registration.

11. CAC released progress on the “Clear and Bright· Addressing Malicious Marketing Chaos in the Short Video Field” special action, notifying the second batch of typical cases (17 July)

The CAC continued to carry out the “Clear and Bright · Addressing Malicious Marketing Chaos in the Short Video Field” special action, guiding and urging short video platforms and platforms providing short video functions to handle illegal and non-compliant accounts in accordance with laws and agreements, and continuously purifying the short video content ecosystem. Typical violations include: fabricating scripts for fake staging and using public goodwill for non-compliant marketing; spreading false information and maliciously disrupting social order; fabricating false personas and exaggerating experiences and qualifications for traffic attraction and marketing; impersonating others' identities for fan attraction and traffic marketing; releasing content violating public order and good customs for bottomless traffic attraction; using exaggerated and false titles and copy for eye-catching, fan attraction, and traffic. The aforementioned accounts have been closed or muted in accordance with laws and agreements.

12. CAC deployed multiple actions to address the online environment for minors, “self-media” publishing false information, and the online environment during the Chengdu World Games (15 July, 29 July, 31 July)

The CAC successively deployed three “Clear and Bright” special actions, aiming to purify the online ecosystem during key periods and scenarios, and prevent online risks involving minors, false information, and major events. First, a 2-month summer online environment remediation for minors was carried out, focusing on four issues in accordance with the Regulations on the Protection of Minors Online: implementing online infringement behaviours, covertly spreading illegal and harmful information, inducing participation in offline dangerous activities, and using minors' images for profit. Second, starting from July 24, a 2-month action to address “self-media” publishing false information was carried out, focusing on issues such as malicious hyping to mislead the public, various means to distort facts, failing to label to confuse real with fake, and false information in professional fields. Third, starting from August 1, a 20-day online environment remediation around the Chengdu World Games was carried out, focusing on handling event-related rumours, group confrontations, impersonating official accounts, leaking athletes' privacy, and online violence, and setting up a reporting zone for rapid handling and collaborative governance.

13. SAMR announced 7 typical cases in the live e-commerce field to guide live e-commerce operators to operate in compliance with the law (8 July)

The SAMR announced 7 typical cases in the live e-commerce field, involving major violations in the e-commerce sector such as false promotion, fake traffic boosting, non-standard promotions, lack of license display, and sales of substandard products: A certain company used false personas for product endorsements and promised “full refund if tuition not recouped” but failed to deliver, constituting false promotion, fined 2 million yuan with license revocation; a certain company failed to display business license on the account homepage and made misleading promotions on product efficacy, fined 400,000 yuan; a certain company conducted improper prize promotions with “payment ranking for free/half-price,” and excluded consumer rights via standard clauses, fined and confiscated 117,900 yuan; an individual “boosted views” for multiple live streams, constituting unfair competition with fabricated data, fined and confiscated 184,500 yuan; a certain company claimed “No.1 sales across the web” in live streams without basis, constituting false or misleading promotion, fined 280,000 yuan; an individual made false promotions on down quilt material, origin, and enterprise honours, fined 200,000 yuan; two certain enterprises produced and sold children's toys not meeting national standards, fined and confiscated 4,877.6 yuan and 25,600 yuan respectively.

14. China Association of Automobile Manufacturers notified testing results on 5 security requirements for automobile data processing, with 49 models from 13 enterprises meeting the requirements (18 August)

The China Association of Automobile Manufacturers issued the third batch notification on testing of “5 Security Requirements for Automobile Data Processing,” aiming to implement the Several Provisions on Automobile Data Security Management (Trial), GB/T 41871-2022, and GB/T 44464-2024, continuously standardize automobile enterprises' data processing, and safeguard user rights. This batch, based on voluntary enterprise submissions, tested five requirements: anonymization of external facial recognition, local processing of cockpit data, default non-collection of cockpit data, prominent notification of personal information processing, and applicability of positioning accuracy, with 49 models from 13 automobile enterprises meeting the standards (some models not involving facial anonymization, meeting the other four). At the same time, the notification pointed out compliance shortcomings in some models: insufficient detection rate for vehicle-end image anonymization (below 90%), setting authorization periods for sensitive personal information as “always allow/permanent,” lacking obvious prompts on the vehicle machine for continuous collection of sensitive information, missing privacy policies in applications, etc., which have been fed back to enterprises for rectification.

15. Beijing CAC issued announcement on filed generative artificial intelligence services, adding 2 generative artificial intelligence services that have completed filing (23 July)

The Beijing CAC issued the announcement on registered generative artificial intelligence services, with Beijing adding 2 generative artificial intelligence services that have completed registration as of July 23, cumulatively completing registration for 144 generative artificial intelligence services.

16. Beijing CAC disclosed algorithm rule principles of major local online platforms, further strengthening algorithm governance and enhancing algorithm transparency (15 August)

The Beijing CAC promoted major local online platforms to disclose the basic principles, purposes, main operating mechanisms, etc., of algorithm recommendation services, collecting and summarizing the algorithm rule principles of major local online platforms, aiming to safeguard users' right to know. Currently, the first batch of 6 platforms have participated in the disclosure, with subsequent disclosures to be made in the “Authoritative Releases” — “Platform Algorithm Principle Disclosure” column on the “CAC Beijing” WeChat official account.

17. Shanghai CAC issued announcement on registered generative artificial intelligence services, adding 4 generative artificial intelligence services that have completed registration (19 August)

The Shanghai CAC issued the announcement on registered generative artificial intelligence services, with Shanghai adding 4 generative artificial intelligence services that have completed registration as of August 19, cumulatively completing registration for 105 generative artificial intelligence services.

18. Guangdong CAC issued second-quarter notification on online reporting handling and enforcement, with one non-compliant AI application delisted and penalized for generated content issues (24 July)

The Guangdong CAC issued the 2025 second-quarter notification on online reporting handling and enforcement, interviewing 187 website platforms in accordance with the law, warning 19, imposing fines on 12, delisting 42 mobile internet applications, and jointly with the provincial communications administration and other departments handling 27 illegal and non-compliant websites. Typical cases include: A certain AI app generated vulgar content, with inadequate review mechanisms, failing to establish network security management systems and emergency plans to prevent the spread of harmful information; a certain website impersonated an enterprise's official site, infringing on enterprise trademark rights and interfering with normal operations; a certain WeChat official account impersonated news media names and logos, illegally conducting internet news information services; a certain WeChat official account released a large amount of illegal and harmful information involving online violence against minors, privacy exposure, attacks and insults, etc., and was closed in accordance with laws and agreements.

19. Hainan CAC advanced online enforcement work, carrying out the “Clear and Bright Free Trade Port” special action (18 July)

The Hainan CAC notified online enforcement achievements since the beginning of the year: Continuing enforcement around information content security, network operation security, data security, and personal information protection, advancing the “Clear and Bright Free Trade Port” special action, cumulatively cleaning up over 13,600 pieces of illegal and harmful information, handling over 32,000 accounts, deregistering and closing 981 websites, interviewing 31 platforms and notifying rectifications 20 times. Among them, typical violations include: Conducting internet news information services without permission, interviewed and fined 10,000 yuan; platforms failing to fulfil primary responsibilities for content security leading to the spread of obscene information, fined 400,000 yuan (including additional fines for overdue); a certain app's youth mode spreading illegal information, fined 200,000 yuan (including additional fines for overdue). In addition, in privacy compliance, the Hainan CAC jointly with relevant departments issued compliance guidelines for personal information protection in the supermarket sector, specially inspected and required rectification within a deadline for 36 illegal apps; in network and data security, the Hainan CAC ordered correction, warned, and fined 50,000 yuan to enterprises failing to fulfil network and data security obligations leading to data leaks; in the financial industry, the Hainan CAC jointly with financial regulatory departments interviewed and corrected financial institutions for excessive collection of personal information.

20. Hangzhou SAMR issued 5 typical cases in the field of internet advertising and promotion to maintain fair competition market order (1 July)

The Hangzhou SAMR announced 5 typical cases of online unfair competition, focusing on chaos such as “moving shop theft” data scraping, technical interference with operations, fake order boosting, name confusion “free riding on famous brands,” and live false scenes, clarifying regulatory scales and compliance boundaries. Typical cases include: A certain software enterprise scraped platform and merchant product data without consent and one-click transferred across platforms, constituting substantial substitution of others' online products or services, fined 530,000 yuan; a certain brand maintenance agency frequently placed bulk orders and returned goods for price control, causing related shops freight losses, inventory backlog, etc., economic losses, and shop downgrades and reputation damage, fined 200,000 yuan; a certain company built a fake order platform to organize false transactions and charge commissions, fined 390,000 yuan; a certain e-commerce enterprise used highly similar logos to well-known software in app names, leading many users to mistakenly believe the two apps were associated, fined 500,000 yuan; a certain jewelry enterprise falsely claimed “first-hand sources” and transaction atmosphere during live sales to mislead consumers, fined 200,000 yuan.

21. Beijing Internet Court concluded case involving AI “deepfake” of celebrity voices for product endorsements, clarifying the scope of natural persons' voice rights protection and merchants' reasonable review obligations (20 August)

The Beijing Internet Court issued a judgment, clarifying that using AI “deepfake” of celebrity voices for product endorsements constitutes infringement of personality rights. The court held that as long as AI-synthesized voices are identifiable in timbre, intonation, pronunciation style, etc., they fall within the scope of the rights holder's voice rights protection; at the same time, extensive use of celebrity portraits paired with synthesized voices for unauthorized commercial promotion constitutes infringement of portrait rights and voice rights. Regarding liability allocation, merchants entrusting “influencers” to release endorsement videos jointly seek benefits for product promotion, and possessing video review and management capabilities, failing to fulfil reasonable review obligations, should bear joint liability with the publisher. The court ordered the involved merchants to apologize and compensate for economic losses and reasonable rights defence expenses totalling 120,000 yuan. This case provides compliance guidelines for e-commerce marketing and deep synthesis applications: Strengthen identification and authorization reviews of synthesized content, improve explicit labelling mechanisms, with merchants and platforms both emphasizing primary responsibilities to prevent misleading consumers and infringing personality rights using AI technology.

Industry Developments

22. State Council issued opinions requiring in-depth implementation of the “AI+” action to promote wide and deep integration of artificial intelligence with various industries and fields of the economy and society (26 August)

The State Council issued the Opinions on Deeply Implementing the “AI+” Action, aiming to promote deep integration of artificial intelligence with the economy and society and cultivate new quality productive forces. The opinions specify achieving large-scale applications in six major areas—science and technology, industry, consumption, livelihood, governance, and global cooperation—by 2027, with the penetration rate of new-generation intelligent terminals/intelligent agents exceeding 70%; by 2030 exceeding 90%, forming an important growth pole for the intelligent economy. The plan first requires accelerating the implementation of key actions: Around “AI+ Science and Technology” to build scientific large models and promote intelligentization of scientific research facilities; “AI+ Industry” to cultivate intelligent native formats, promote full-process industrialization and digital intelligence in agriculture, and intelligentization in services; “AI+ Consumption” to expand intelligent services and intelligent terminal ecosystems; “AI+ Livelihood” to promote human-machine collaboration in scenarios such as education, healthcare, and elderly care; “AI+ Governance” to promote urban and government intelligentization, public safety, and ecological governance applications; “AI+ Global Cooperation” to promote open-source universality and global governance coordination. Second, the plan proposes strengthening basic support capabilities: Enhancing model basic capabilities and evaluation systems, strengthening data supply innovation, coordinating intelligent computing power, optimizing application development environments, promoting prosperous open-source ecosystems, and strengthening talent team building. Finally, in policy and security aspects, the plan requires improving laws, regulations, and ethical norms, optimizing security assessments and filings, building full-chain security capabilities from models, data, infrastructure to applications, and monitoring and early warning—emergency response mechanisms, implementing inclusive, prudent, classified, and graded governance.

23. World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance issued plan to guide global governance actions for artificial intelligence (26 July)

The World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance issued the Action Plan for Global Governance of Artificial Intelligence, aiming to create an inclusive, open, sustainable, fair, secure, and reliable digital and intelligent future. The plan calls on all parties to take effective actions to jointly advance global artificial intelligence development and governance, specifically: Jointly seizing artificial intelligence opportunities; promoting innovative development of artificial intelligence; advancing artificial intelligence empowerment across industries; accelerating digital infrastructure construction; creating diverse, open, and innovative ecosystems; actively promoting high-quality data supply; effectively addressing energy and environmental issues; promoting consensus on standards and norms; public sectors taking the lead in deployment and application; conducting artificial intelligence security governance; jointly implementing the Global Digital Compact; strengthening international cooperation in artificial intelligence capacity building; building an inclusive governance model with multi-party participation.

24. Shanghai Municipal Government Office issued implementation plan to guide the development of embodied intelligence industries, implementing the national “AI+” action deployment (6 August)

The Shanghai Municipal Government Office issued the Implementation Plan for the Development of Embodied Intelligence Industries in Shanghai, aiming to implement the “AI+” deployment and build a highland for embodied intelligence industries. The overall goals are to achieve breakthroughs in 20 core algorithms and technologies by 2027, build ≥4 incubators, land “100 scenarios” and “100 products,” with core industry scale exceeding 50 billion yuan. In model innovation drive aspects, the plan will focus on supporting breakthroughs in perception decision models, motion control models, embodied collaboration technologies, and corpus technologies, while supporting the creation of autonomous operating systems. In public platform aspects, the plan supports computing power platforms, digital twin training fields, pilot test platforms, investment platforms, and leasing platforms with specific measures. In application demonstration benchmarks, the plan supports embodied intelligence in logistics assembly, industrial manufacturing, commercial retail, medical health and wellness, and home services. In cluster chain collaborative development, the plan supports developing core components, embodied intelligence terminal products, optimizing and strengthening high-quality enterprises, and building characteristic industry clusters. In shaping a first-class ecosystem, the plan intends to strengthen open-source innovation, improve standard systems, ensure security ethics, accelerate talent guidance, and deepen open cooperation.

25. Shanghai EITC issued several measures to further expand artificial intelligence applications and accelerate the implementation of the “AI+” action (28 July)

The Shanghai EITC issued the Several Measures of Shanghai for Further Expanding Artificial Intelligence Applications, aiming to further expand artificial intelligence applications and accelerate the building of a world-class artificial intelligence industry cluster. Specific measures include: Reducing the cost of using intelligent computing power, expanding applications of artificial intelligence large models, supporting the procurement of high-quality corpora, carrying out key technology innovations, promoting changes in scientific research paradigms, building industry innovation service platforms, accelerating the promotion and application of innovative products, opening up and constructing benchmark demonstration scenarios, accelerating industry cluster development, strengthening artificial intelligence talent services, improving industry investment and financing systems, and building an open-source and open innovation ecosystem.

26. Shanghai EITC issued implementation plan to accelerate the promotion of “AI+ Manufacturing” development (19 August)

The Shanghai EITC issued the Implementation Plan of Shanghai for Accelerating the Promotion of “AI+ Manufacturing” Development, aiming to implement the “Molding Shanghai” project and accelerate the intelligent development of manufacturing. The implementation plan expects to promote intelligent applications in 3,000 manufacturing enterprises over 3 years; build 10 industry benchmark models, form 100 benchmark intelligent products; promote 100 demonstration application scenarios, build about 10 “AI+ Manufacturing” demonstration factories; develop about 5 comprehensive integrated service providers. In tackling basic and frontier technologies, the implementation plan intends to enhance industrial model basic capabilities, breakthrough frontier technologies in industrial intelligence, and develop industrial data governance and synthesis technologies. In building key element platforms, the implementation plan intends to build industrial intelligent computing cloud platforms, industrial corpus public service platforms, and integrated innovation bases. In promoting applications in key industries, the implementation plan expects to accelerate empowerment in key industries such as circuits, electronic information, automobiles, and high-end equipment, build common demonstration scenarios, explore frontier manufacturing modes, and build “AI+ Manufacturing” factories. In building “AI+” industrial intelligent products, the implementation plan emphasizes AI+ industrial software tools, industrial products and equipment, and consumer terminals. In actively creating a development ecosystem, the implementation plan requires cultivating professional service providers, promoting intelligent upgrades of platforms, accelerating robot applications, strengthening scenario guidance, cultivating professional talents, promoting industry-finance docking, and improving standards and security systems.

27. Internet Society of China issued guidelines to provide useful guidance for compliance management of user rights protection in mobile internet application services (4 July)

The Internet Society of China issued the Compliance Management Guidelines for User Rights Protection in Mobile Internet Application Services (2025), aiming to provide operable user rights protection compliance frameworks for entities such as apps, mini-programs, SDKs, application distribution platforms, and intelligent terminals. Based on sorting out current regulations, the guidelines propose clear requirements around six aspects: service provision, personal information protection, algorithm recommendations, charging management, complaint handling, and customer service; simultaneously proposing three types of compliance management mechanisms: First, organizational mechanisms, clarifying responsibilities of boards, executives, and leading departments, refining business divisions; second, operational mechanisms, strengthening system construction, risk identification and assessment, incident emergency and handling, internal audits, and rectification closed loops; third, assurance mechanisms, covering personnel allocation and authority management, full-staff training, performance assessments and rewards/punishments, and digital support. The guidelines also list applicable objects, term explanations, and regulation lists, facilitating enterprises' benchmarking execution, conducting promotion training, and self-inspection rectifications.

28. 18 enterprises including Alibaba and Baidu disclosed practical results of the Artificial Intelligence Safety Commitments to promote new progress in AI security governance (18 July)

The China Artificial Intelligence Industry Development Alliance released the Disclosure of Practical Results of the Artificial Intelligence Safety Commitments. The disclosure of practical results revolves around six major commitment dimensions: “risk management, model security, data security, infrastructure security, transparency, and frontier security research,” extracting 20 key security labels (such as security organizations and responsibilities, baselines and assessments, red team testing, emergency responses, external disclosures, etc.), publicly sharing 43 typical practices, forming a benchmarkable and reusable compliance and technical practice library. Currently, 22 enterprises have signed the commitments so far, with 18 actively disclosing measures.