China is speeding up a comprehensive artificial intelligence law that would govern AI models from development through deployment, with the country's top economic planning agency naming frontier developers DeepSeek, Alibaba and Moonshot as squarely in scope. The announcement, made on Tuesday, 3 November 2026, by the National Development and Reform Commission (NDRC), marks one of Beijing's clearest moves yet towards a single, full-lifecycle rulebook for the technology reshaping its economy and its rivalry with the United States.

Senior NDRC official Wang Ruomeng said the legislation would cover the core enablers of AI — data, computing power and algorithms — and key application scenarios. Developers and operators will be expected to strengthen risk management and disclose safety risks, incidents and vulnerabilities, with regular reviews to update outdated provisions.

Key Points

• China's NDRC said on 3 November 2026 it will accelerate comprehensive AI legislation and safety standards covering the full lifecycle of models.
• Senior official Wang Ruomeng named DeepSeek, Moonshot and Alibaba Group as developers in scope, describing them as leading China's effort to rival OpenAI and Anthropic.
• The law will govern data, computing power and algorithms, and will require risk management plus disclosure of safety risks, incidents and vulnerabilities.
• Beijing plans sector-specific regulatory sandboxes, an unspecified transition period and compliance guidance, alongside monitoring of AI's impact on jobs.
• The move follows a September China-US agreement on AI's benefits and threats, and AI discussions at the recent Trump-Xi summit in Washington.

What Happened

The NDRC's announcement came as China steps up AI policymaking. Rather than narrow rules for one kind of model or industry, officials described a comprehensive law with safety standards and ethical guidelines applying across the whole lifecycle of an AI system — from training data and computing power through to deployment to users.

Wang placed three domestic names at the centre: DeepSeek, the start-up whose low-cost models stunned global markets; Moonshot, another fast-rising frontier developer; and Alibaba Group, the e-commerce and cloud giant now pouring resources into AI. He described them as leading China's attempt to rival OpenAI and Anthropic.

For developers and operators, the message was accountability. Companies will be expected to bolster risk management and disclose safety risks, incidents and vulnerabilities rather than handling them quietly, and implementation will be reviewed regularly.

Background/Context

China has regulated parts of AI for several years, with rules covering recommendation algorithms, deep synthesis and generative AI services. What has been missing, analysts have long noted, is a single overarching law tying those pieces together with clear duties across a model's entire lifecycle.

The timing matters. The announcement follows a September China-US agreement on a common understanding of AI's benefits and threats, and comes after AI was discussed at the recent Trump-Xi summit in Washington. It also sits inside China's 15th five-year plan, running to 2030, which still encourages open AI development as regulation tightens.

That dual track — encouragement plus control — has defined Beijing's approach: domestic champions strong enough to compete globally, while risks from misinformation to labour disruption stay under state supervision. A full-lifecycle law would give regulators one framework to pursue both aims.

China AI Law: What the Full-Lifecycle Rules Would Cover

At the heart of the plan are the three enablers officials named: data, computing power and algorithms. Data rules would shape what models can be trained on and how sensitive information is handled. Computing-power provisions touch the chips and data centres that are a strategic bottleneck, particularly as US export controls limit China's access to the most advanced hardware. Algorithm rules would govern how models are designed, tested and updated.

Beyond those enablers, the law is expected to reach key application scenarios — healthcare, finance, education and public services — where AI decisions affect people's lives. Sector-specific regulatory sandboxes would let companies trial new systems under supervision, so regulators can spot problems before a product reaches millions of users.

Officials also promised an unspecified transition period and compliance guidance, acknowledging even large developers need time to adapt. Beijing also plans to monitor AI's impact on employment and build safety-net protections for displaced workers.

Reactions/Quotes

Wang framed the effort as one no country can manage alone. "AI safety is not a problem one company, industry or country can tackle alone," he said, calling for international cooperation on vulnerability disclosure and risk responses — a recurring theme in Beijing's AI diplomacy.

The NDRC published no draft text or timetable, and gave no date for the law taking effect. Enforcement powers, penalties and the precise scope of disclosure duties will be settled in drafting, where industry lobbying and inter-agency negotiation are likely to be intense.

For the named developers, the announcement cuts both ways. Being singled out as national champions confirms their standing; being placed in scope of a comprehensive law means their safety practices and incident reporting will face closer scrutiny.

Why It Matters

A full-lifecycle AI law from China would matter well beyond its borders. China is home to some of the world's most widely used AI models, and rules set in Beijing shape how those models are built, what data they use and what safeguards they carry when exported or offered online worldwide.

It also matters for the US-China technology contest. Washington has pursued export controls, investment screening and voluntary safety commitments, while the European Union has adopted a risk-based AI Act. China's law would add a third major model, with the state playing a more direct role.

For workers, the employment provisions may prove as important as the technical rules. Pairing the law with job-displacement monitoring and safety-net protections signals Beijing sees AI as a force that could reshape labour markets within the current five-year plan.

What Happens Next

Next comes drafting and consultation: a formal bill, safety standards, ethical guidelines and sandbox designs, plus clarity on the transition period and how existing sector rules fit inside the new framework.

Watch too for how the law interacts with the September China-US understanding on AI and any follow-up from the Trump-Xi summit. Even loose agreement on vulnerability disclosure could move the global safety debate towards shared practice.

For DeepSeek, Moonshot, Alibaba and their rivals, the immediate task is preparation: auditing data pipelines, documenting risk controls and building incident-reporting systems the law is expected to demand.

Frequently Asked Questions

What is China's new AI law? A planned comprehensive law, announced by the NDRC on 3 November 2026, governing AI models across their full lifecycle, with safety standards and ethical guidelines.

Which companies are in scope? Wang Ruomeng named DeepSeek, Moonshot and Alibaba Group as leading domestic frontier developers in scope, describing them as rivals to OpenAI and Anthropic.

What will companies have to disclose? Safety risks, incidents and vulnerabilities, with stronger risk management and regular reviews of implementation.

What are the regulatory sandboxes? Sector-specific spaces to trial AI systems under supervision, letting regulators test rules and spot problems before wide deployment.

When will the law take effect? No date has been announced; officials promised an unspecified transition period and compliance guidance, with drafting still to come.

Read more on US News Pulse: Trump backs Nvidia CEO Jensen Huang amid reported China spy plot inquiry and Hidden deepfake apps exposed: 400 ordinary AI apps concealed abusive tools.