COMMENTARY & ANALYSIS

Commentary: As AI Risks Rise, China and U.S. Search for Different Guardrails

By Lan Jie and He Yujie

Published: Aug. 20, 2026 12:44 p.m. GMT+8

00:00

00:00/00:00

Listen to this article 1x

China and the U.S. are pursuing different approaches to AI governance. Photo: IC photo

Artificial intelligence is moving from answering questions to taking action. The rise of AI agents capable of using tools, executing commands and making decisions has expanded both the technology’s usefulness and the potential consequences when it is misused or fails.

In a February 2026 report on international AI safety, Turing Award laureate Yoshua Bengio groups the risks from advanced AI into three broad categories: malicious use, malfunction and systemic risks.

You've accessed an article available only to subscribers

Subscribe today for just $.99.

VIEW OPTIONS

Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.

Subscribe to both Caixin Global and Bloomberg - for the price of one.

Share this article

Open WeChat and scan the QR code

DIGEST HUB

Digest Hub

Back

Explore the story in 30 seconds

Yoshua Bengio categorized AI risks into malicious use, malfunction, and systemic risks.
China employs mandatory filings and safety assessments; US relies on voluntary rules and state regulations.
Both share concerns over misuse and agreed to resume AI dialogue in May 2026.

AI generated, for reference only

Explore the story in 3 minutes

1. The article discusses the evolution of AI into autonomous agents capable of using tools and making decisions, expanding both its utility and the potential consequences of misuse or failure [para. 1]. Yoshua Bengio groups the risks of advanced AI into malicious use (cyberattacks, fraud, weapon assistance), malfunction (errors from capability limits or misaligned objectives), and systemic risks (impacts on employment, intellectual property, and information ecosystems) [para. 2][para. 3]. Policymakers face a dilemma: they cannot wait for major damage before responding, but must avoid constraining innovation by treating every uncertainty as an immediate threat [para. 4].

2. China’s governance strategy emphasizes parallel development of industry and safety, beginning with its 2017 State Council plan [para. 5][para. 6]. It has built a layered regulatory system grounded in foundational laws, supplemented by specific rules for algorithms, generative AI, and anthropomorphic services [para. 7]. Termed “agile governance” by Tsinghua’s Xue Lan, the framework relies on algorithm filings, mandatory safety assessments, content labeling (codified in a 2025 national standard), and ethical review procedures [para. 8][para. 9][para. 10][para. 11][para. 12][para. 13]. While effective against malicious use and technical failures, China’s current tools are less developed for resolving deeper systemic issues like employment disruption, requiring closer integration with civil and labor law [para. 14][para. 15].

3. The U.S. approach under the Trump administration sharply prioritizes innovation and global leadership with minimal federal regulatory burdens [para. 16][para. 17]. Federal policy, outlined in executive orders and the 2025 AI Action Plan, focuses on accelerating innovation and limiting state regulation, with a 2026 White House framework addressing child protection and IP while maintaining a pro-innovation stance [para. 18][para. 19][para. 20]. Crucially, the federal approach relies on voluntary cooperation with frontier developers on cybersecurity, explicitly avoiding the mandatory filing, licensing, and pre-release approval mechanisms seen in China [para. 21][para. 22]. In contrast, states actively regulate specific risks: many regulate chatbots for minors and mental health, while California and New York have enacted laws requiring binding safety frameworks for frontier models [para. 23][para. 24][para. 25]. This creates a distinct dual system of federal innovation promotion and state-level risk regulation [para. 26].

4. Despite their differing philosophies, both countries face the challenge of technology advancing faster than regulation [para. 27]. They share concerns over model proliferation, cyberattacks, and misuse by nonstate actors, creating room for technical cooperation [para. 28]. The agreement during President Trump’s May 2026 visit to China to resume AI dialogue suggests that competition and communication on risk safeguards can coexist, which is vital for global stability [para. 29].

AI generated, for reference only

What Happened When

2017：: The State Council’s New Generation Artificial Intelligence Development Plan identified AI as a strategic technology while warning of risks to employment, privacy, ethics, economic security and social stability.

By July 2025：: The Trump administration followed with an AI Action Plan centered on accelerating innovation, expanding infrastructure and strengthening U.S. leadership abroad.

As of 2025：: Regulations and a mandatory national standard for identifying AI-generated content took effect in China, establishing requirements for visible and embedded labels across text, images, audio, video and virtual content.

2026：: Measures for ethical review of AI research and development were introduced in China, setting out procedures for reviewing risks involving human dignity, public order, health, the environment and sustainable development.

AI generated, for reference only

GALLERY

: Gallery: Zhu Rongji’s Legacy

PODCAST

China Business Uncovered Podcast: Leaked Deals, Bribes Behind Hong Kong’s Insider-Trading Scheme

00:00

00:00/00:00

MOST POPULAR

1: Crypto Investor Harry Yeh Dies in Mysterious Plunge From Paraguay High-Rise

2: Commentary: The Real Flaw in China’s AI Strategy

3: Commentary: The Phantom Menace of Chinese Overcapacity

4: China to Add Petrochemicals, Chemicals to National Carbon Market

5: DeepSeek Launches V4-Pro and Raises API Prices by as Much as 1,100%