Part one of this series covered how China governs artificial intelligence: no single law, the Cyberspace Administration of China as lead watchdog, and a stack of narrow rules that grows one entry at a time. That answered the “how.” This piece answers the harder question: now that we know how Beijing does it, should Europe be worried?
The popular story says China moves fast because its rules are flexible, while Europe drags itself down with one giant law. That was roughly true two years ago. In 2026, both sides are walking toward each other, and the real risk for European businesses lies elsewhere.
Two Opposite Bets: One Big Law vs. Dozens of Small Ones
Europe made one bet. The EU AI Act is a single, horizontal law that sorts AI by risk. It regulates the technology itself and leans on protecting rights with documentation, human oversight, and conformity checks for high-risk systems.
China made the opposite bet. Beijing governs through targeted rules aimed at specific uses: recommendation algorithms got one rulebook, deepfakes another, generative AI a third, and content labeling a fourth. The focus is on what AI produces and how it is used, with content control and social stability as the organizing concerns.
The Same Chatbot Faces Different Gates in Brussels and Beijing
To serve the public in China, a generative AI assistant must file with regulators if it can sway public opinion, pass a security assessment, train on lawfully sourced data, and keep its outputs inside “socialist core values.” Any AI system operating in China must comply with content moderation rules that are incompatible with how it operates in the EU. That is not a setting you flip; it means different model behavior, training data, and output filtering.
The same tool entering Europe meets different gates: transparency duties so users know they are talking to a machine, copyright and data rules on training material, and a full high-risk compliance file if it lands in a sensitive area like hiring or credit.
DeepSeek is the live example. The Chinese model spread fast across Europe in early 2025, then hit a wall. Italy’s regulator pulled its app over data-transfer transparency failures, and Germany’s Berlin commissioner declared its practices unlawful under GDPR, asking Apple and Google to drop it from their stores. The commissioner found the company could not show that German users’ data was protected in China at a level equivalent to EU standards. A product built to clear Beijing’s gates could not clear Brussels’.
China’s Speed Comes With a Price Tag
China’s piecemeal approach moves quickly. When ChatGPT sparked a global scramble, China’s generative AI measures took effect in August 2023, marking the first regulation built for the technology. Writing a narrow rule, watching the market, then writing the next one lets regulators react in months. Europe’s response to ChatGPT, by contrast, forced AI Act drafters to bolt on new provisions mid-negotiation.
But speed buys uncertainty. China’s fast-evolving rules are not yet fully stabilized, leaving companies with gray areas in compliance. A rulebook that grows every quarter is one you can never finish reading. Lawyers tracking both systems note that flexible regimes leave businesses unsure how obligations will be read later, which can make long-term AI investment harder to attract. Europe’s one law is slower and heavier, but it gives a company a single text to plan against for years.
The Two Systems Are Swapping Playbooks in 2026
Europe spent 2026 making its single law lighter and slower. On 7 May 2026, the Parliament and Council reached a provisional deal on the Digital Omnibus on AI, the first amendments since the Act was passed. High-risk obligations for stand-alone Annex III systems were deferred to 2 December 2027, and for AI embedded in regulated products under Annex I to 2 August 2028. The package also trimmed duplication for industrial AI and eased rules for smaller firms. The trigger was that by late 2025, implementation was visibly off track.
China moved the other way. On 11 May 2026, the State Council included comprehensive AI legislation in its annual work plan in unusually direct language, the first time authorities had described planned AI legislation in such detail. The country built on having no single AI law is now drafting its first unified one.
Read together, the neat contrast collapses. Europe is loosening its big law to move faster. China is reaching for a big law to add the structure its patchwork lacks. The systems are converging from opposite ends.
So Should European Business Worry? Yes, but Not About the Rulebook
The EU is behind on nearly every key AI metric except regulation. In 2024, U.S. institutions produced 40 notable AI models to China’s 15 and Europe’s three, according to Stanford’s AI Index. Its 2026 edition finds model production still concentrated in the U.S. and China, with the two trading the frontier lead repeatedly through 2025.
This month Beijing set out plans to spend roughly 2 trillion yuan (about $295 billion) over five years on a state-directed, nationwide AI data-centre buildout, designed to run on domestic chips and largely lock out Nvidia and AMD.
Europe’s problem sits upstream of any rulebook. Its financial system is bank-centred (bank assets run to roughly 300% of EU GDP against about 85% in the US), which leaves far thinner venture capital for the high-risk, capital-intensive bets frontier AI demands, and steers what capital exists toward incumbents over startups.
So aim the worry correctly. China’s regulation strategy is not the thing eating Europe’s lead; capital, compute, and the cost of scaling are problems Brussels was already chasing through the European Innovation Act. What China’s approach gives Beijing is tempo: the ability to write a rule for a new use within months and adjust as the market reacts. That tempo, paired with money and chips Europe cannot yet match, is the edge worth watching, not the absence of an “AI Act.”
A company can comply with any rulebook given lawyers and lead time. It cannot conjure venture capital, foundation models, and chips out of a compliance manual. The rulebooks are converging. The capital is not.
Author: Ayanfe Fakunle
See Also:
How Does China Regulate AI in 2026? Beijing’s Rulebook Without a Single Comprehensive Law
