Policy desk

Automakers can apply for the second batch of the L3 access pilot to obtain conditional approval

Compute capacity in Germany will expand significantly by 2030, giving companies deploying AI training or inference in Germany more local compute supply

M- and N-category vehicles with L3/L4 autonomous systems must comply with the mandatory standard before 1 July 2027 or cannot be sold.

Federal civilian agencies and their suppliers cannot deploy or retain DeepSeek products or third-party services built on them in government IT systems.

Federal civilian agencies using DeepSeek models or services must stop and switch to alternatives.

Generative AI medical device developers can submit feedback during the open docket period and directly shape the future regulatory framework.

Mountain scenic areas in Shandong present procurement and deployment opportunities for hiking-assist robot vendors

Vendors serving US federal agencies must audit and remove DeepSeek technology or derivative components integrated in their products.- Third-party services built on DeepSeek (fine-tuned, distilled, or packaged) are also covered; companies with US government business need to plan model replacement.- Chinese AI models face tighter access to the US federal market; DeepSeek-dependent go-to-market strategies need reassessment.

Companies selling to US federal agencies or their suppliers must verify their tech stack contains no DeepSeek or derivative models.

Vendors and integrators in federal civilian supply chains must identify and replace products built on DeepSeek.- Companies building on Chinese open-source models face procurement restrictions when targeting US government customers.- Tightening two-way controls between the US and China raises compliance scrutiny for Chinese AI firms operating in the US and vendors with ASEAN footprints.

Procurement of DeepSeek-origin or DeepSeek-architecture-based hardware, software, and services to U.S. federal civilian agencies is prohibited, blocking related contracts immediately.

Companies selling to US federal agencies must confirm their supply chains contain no DeepSeek components or risk losing procurement eligibility

Federal software vendors must audit which open-weight models are embedded in products sold to civilian agencies and demonstrate supply chain compliance to keep contracts

Models exceeding 2T parameters or 10^26 FLOPS training compute will require export approval, and public open-source releases could be restricted — release strategies must be re-evaluated

Providers of high-risk AI systems that implement EN 18286 gain a presumption of conformity with Article 17 QMS requirements

Commerce and manufacturing companies can apply for inclusive computing power, 'specialised and sophisticated' enterprise status, and fiscal support under the 'AI + Manufacturing' action.

Transfers of AI technology, models or related technical materials abroad (including to ASEAN) will require prior export licences, raising deal timelines and compliance costs

Exporting non-public large model, agent, or computing cluster technologies via licensing, sale, or weight delivery requires interim registration with MOFCOM during the transition; in-flight transactions are also covered, with no grandfathering.

Export of large models, AI agents and computing cluster technologies would become restricted, so cross-border licensing, partnerships and deployments need export-control review

Export of model weights, source code, and training methods above the 10^26 FLOPs or 2 trillion parameter thresholds will require approval before release, delaying or blocking cross-border collaborations

Cross-border export or transfer of AI, biopharma or solar technology will require approval, so companies must plan licensing ahead

Chinese companies transferring AI models, algorithms or related technology abroad will face new export licensing requirements.

Companies exporting large model, agent, or computing cluster technologies must complete export licensing or pre-export registration for the 23 AI items

Overseas localised deployment of restricted large models falls under export control and requires a licence (automatic licence available for deployment on overseas self-owned cloud); API calls, SaaS apps and lightweight on-device deployment are outside scope.

Foreign AI services must obtain Chinese AI approval before launching in China; Apple's approval provides a reference path for similar products.

Companies deploying AI agents in Singapore and the region can use the reference architecture to design inter-agent authentication and trust transfer, improving interoperability.

Licensed banks operating in Hong Kong must establish clear accountability mechanisms and ownership for AI systems under the ART framework

Frontier model providers operating in the EU face AI Act obligations on safety-incident reporting and cooperation with investigations; insufficient explanations escalate to formal enforcement.

On-device LLMs offered in China must complete CAC filing before launch; devices without updated filings (e.g. iPhone 15/15 Plus) do not carry the AI features

Companies must use IMDA's testing kit to evaluate agentic AI system safety