Part of the AI Regulation News hub.
APEC member economies issued a statement on AI at a forum hosted by China in Chengdu, committing to facilitate cross-border data flows and to support open-source models that employ strong security assurance
Nothing here binds anyone. What it does is put twenty-one economies on record about two contested questions: whether data should move across borders for AI, and whether open-weight models belong in a security-first policy framework. On both, the answer is yes.
Bottom line: A non-binding statement from a High-Level Forum, not a ministerial decision or an instrument with legal effect. Its value is as a record of consensus positions and as the first commitment to hold AI policy dialogues inside the APEC Digital Economy Steering Group.
Who this affects: Policy and government affairs teams operating across the Asia-Pacific, anyone arguing data-localisation questions in the region, open-source model developers, and trade counsel tracking AI in customs and trade facilitation.
Issue date: 24 July 2026, Chengdu, Sichuan, China.
What changed: No obligation. What is new is the alignment: the statement names cross-border data flows as critical to AI development, and endorses open-source models and projects that employ strong security assurance, subject to security, data protection and intellectual property rights.
Analysis: Read this against the region's actual regulatory direction. Several APEC economies operate or are building data-localisation regimes, and several are tightening rules on model release. A consensus statement pointing the other way does not override any of that, but it is the text a company will cite when arguing that a proposed localisation measure runs against a position its own government has endorsed.
Primary sources: Statement text, Chinese and official English, as published by the host economy
- Instrument
- APEC High-Level Forum on AI, Statement on Promoting AI Development in the Asia-Pacific Region
- Forum
- APEC High-Level Forum on AI, hosted by China under the theme Promoting AI Development for Good and for All
- Jurisdiction
- Asia-Pacific Economic Cooperation member economies
- Status
- Issued 24 July 2026. Non-binding.
- Bindingness
- None. A statement of encouragement and intent. It uses encourage, promote, welcome and commit to organizing; it creates no obligation and no compliance date.
- Instruments referenced
- APEC Putrajaya Vision 2040; APEC Internet and Digital Economy Roadmap (AIDER); APEC AI Initiative (2026-2030)
- Follow-on mechanism
- Commitment to organise policy dialogues on AI during the plenary of the APEC Digital Economy Steering Group, with the Telecommunications and Information Working Group and other APEC sub-fora named
- Sourcing note
- Published by the Cyberspace Administration of China as host, in Chinese with the official English wording on the same page. We did not locate the statement on apec.org.
- Editorial Note
- Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
- Primary source
- https://www.cac.gov.cn/2026-07/24/c_1786639246282926.htm
The two positions worth extracting
Most of this statement is the language of encouragement, and encouragement is not regulation. Two passages are nonetheless worth lifting out, because they take sides on questions that are actively contested in the region.
The first is on data movement. Under the heading of promoting the secure development and deployment of AI, the statement says member economies will work to facilitate cross-border data flows, which are critical to AI development, while strengthening consumer and business trust. That is a consensus characterisation of cross-border transfer as a precondition for AI, not merely as a trade convenience.
The second is on open models. Under fostering an enabling environment for AI innovation, and while respecting security, data protection and intellectual property rights, member economies are encouraged to support open-source models and projects that employ strong security assurance through development and deployment.
Note the qualifier in that sentence. The endorsement is not of open weights as such, but of open-source models that carry security assurance through both development and deployment. That is a narrower proposition than open release, and it is the form in which open models are becoming acceptable in official texts.
What the statement is, institutionally
The forum was held in Chengdu on 24 July 2026, hosted by China, under the theme Promoting AI Development for Good and for All. The statement records that APEC member economies reaffirmed a commitment to deepening exchanges, building consensus, forging cooperation and promoting innovation on AI.
It situates itself inside three existing instruments: the APEC Putrajaya Vision 2040, the APEC Internet and Digital Economy Roadmap, and the APEC AI Initiative for 2026 to 2030. The last of those is the operative programme, and its existence is worth noting on its own for anyone tracking regional AI governance.
The concrete institutional commitment is procedural. Recognising the role of the APEC Digital Economy Steering Group, the Telecommunications and Information Working Group and other sub-fora, member economies commit to organising policy dialogues on AI during the Steering Group plenary, covering AI innovation, adoption, access and other emerging issues.
That is the part with a forward mechanism. Everything else is encouragement; the dialogues are a scheduled venue where positions can harden into work programmes.
The four cooperation areas, briefly
Secure development and deployment. Economies are encouraged to prioritise security, to exchange best practices on reliable, accessible, scalable, trusted, secure and resilient AI infrastructure, and to enhance the accessibility of digital infrastructure, alongside the cross-border data flows commitment.
Responsible adoption. The statement promotes accessible education and training resources across all segments of society to improve AI literacy and skills, and recognises the growing role of AI in facilitating trade, modernising customs procedures and enhancing business competitiveness.
An enabling environment. It commends market-oriented and innovation-driven policies for development and broad adoption, and encourages economies to consider policies that address risks and promote responsible adoption.
Broad-based cooperation. It welcomes capacity building and knowledge sharing across the region through open dialogue and collaborative efforts including with the public and private sectors, and commits to supporting all member economies in participating in the AI-driven economy.
The customs and trade facilitation point is the most concrete of these. AI in customs procedures is an area where APEC work programmes have historically produced practical output rather than statements.
How to use it, and how not to
This is soft law of the softest kind: a forum statement, not a leaders' declaration or a ministerial decision, with no compliance date and no review mechanism beyond the promised dialogues.
It should not be reported as APEC adopting an AI policy, and it does not constrain any member economy. Several APEC economies maintain data-localisation requirements, and nothing here disturbs them.
Its practical use is evidentiary. Where a company or trade association is arguing against a proposed localisation measure or a restriction on model release, this text is a position the relevant government has itself endorsed in a multilateral setting. That is a modest but real argument, and it is why the exact wording matters more than the summary.
One caution on provenance, which we state rather than paper over. China chaired and hosted the forum, and the text used here is the one published by the Cyberspace Administration of China, which carries the Chinese version and the official English wording on the same page. We were not able to locate the statement on apec.org. Readers relying on the wording for advocacy should be aware of that, and may wish to confirm the text against their own economy's published copy.
Two sentences are worth keeping. Member economies will work to facilitate cross-border data flows, which are critical to AI development, while strengthening consumer and business trust. And, while respecting security, data protection and intellectual property rights, they encourage support for open-source models and projects that employ strong security assurance through development and deployment. Neither binds anyone, and neither displaces a single localisation rule now in force. Both are positions your counterpart's government has endorsed multilaterally, which is exactly the kind of text that is useful in a consultation response and useless in a courtroom. Watch the APEC Digital Economy Steering Group plenary, since that is where the statement commits to continuing the conversation.
Source File
https://www.cac.gov.cn/2026-07/24/c_1786639246282926.htm
Open the CAC page and confirm three things: the Chinese headline recording the APEC High-Level Forum statement dated 24 July 2026; the English text beginning APEC High-Level Forum on AI Statement on Promoting AI Development in the Asia-Pacific Region, which appears below the Chinese; and within that English text the sentences on facilitating cross-border data flows and on supporting open-source models that employ strong security assurance.
We will work to facilitate cross-border data flows, which are critical to AI development, while strengthening consumer and business trust. APEC High-Level Forum on AI, Statement on Promoting AI Development in the Asia-Pacific Region, Chengdu, 24 July 2026
FAQ
Is this binding on APEC economies?
No. It is a statement from a High-Level Forum, expressed in the language of encouragement and intent. It sets no obligation, no compliance date and no review mechanism beyond a commitment to hold policy dialogues during the APEC Digital Economy Steering Group plenary.
What does it say about cross-border data flows?
That member economies will work to facilitate cross-border data flows, which are critical to AI development, while strengthening consumer and business trust. It appears under the heading on promoting the secure development and deployment of AI.
What does it say about open-source models?
That while respecting security, data protection and intellectual property rights, member economies are encouraged to support open-source models and projects that employ strong security assurance through development and deployment. The endorsement is qualified by that security assurance condition.
Which existing instruments does it build on?
The APEC Putrajaya Vision 2040, the APEC Internet and Digital Economy Roadmap, and the APEC AI Initiative for 2026 to 2030.
Does it create any follow-on process?
Yes, a procedural one. Member economies commit to organising policy dialogues on AI during the plenary of the APEC Digital Economy Steering Group, recognising also the Telecommunications and Information Working Group and other APEC sub-fora.
Where is the authoritative text?
The version used here was published by the Cyberspace Administration of China, which hosted the forum, and carries the Chinese text with the official English wording on the same page. We did not locate the statement on apec.org, and readers relying on the exact wording may wish to confirm it against their own economy's published copy.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.