On 23 September 2025 the 2025 AI Industry and New Industrialisation Conference opened in Beijing, with more than 500 delegates from industry, academia, research, application and finance attending in person and several landmark results released.
The policy signal
The value of such industry conferences lies less in the agenda than in the shift of policy emphasis they mark. For a period, AI policy focused on building general capability — compute, data and models. This conference put “empowering new industrialisation” in a prominent position, meaning policy resources are tilting towards industrial settings.
That matches the overall design of the “AI plus” initiative: the value of general capability ultimately shows in productivity gains in specific industries. See our notes on the State Council’s “AI plus” deployment and the AI plus manufacturing action plan.
Why industrial settings are deeper compliance water
Industrial AI is more complex than general use for three reasons.
Physical safety responsibility. In general settings, AI errors mostly mean inappropriate content; in industrial settings, they can cause equipment failure, production accidents or personal injury. Liability shifts from content to product and safety responsibility.
Layered data structures. Industrial data includes process parameters, equipment data, supply chain data and personal data, so classification, grading and cross-border flow are significantly more complex. See our notes on the industrial high-quality dataset plan and data element market allocation.
Closed control systems. Industrial control systems have long operated in relatively closed networks; introducing AI changes the security boundary and creates new cyber risk. See our note on the AI clause in the revised Cybersecurity Law.
Directions companies can anticipate
First, AI products for industrial settings need safety assessment extended from the model layer to the system and site layers. Second, when working with manufacturers, liability allocation clauses become central to contract negotiations, especially for accidents. See our note on AI commercial contract clauses. Third, compliant sourcing and authorisation chains for industrial datasets are one of the gates to this market.
The policy window for industrial settings is open, but capturing it depends on whether companies front-load compliance into product design.
This article is based on public information from the 2025 AI Industry and New Industrialisation Conference. It is for general reference only and does not constitute legal advice.