News Reporter |
News Editor | Wen Shuqi
At WAIC 2026, AI, tokens, and agents popped up in nearly every forum and booth discussion. According to the newly released “World Artificial Intelligence Rule of Law Blue Book (2026),” over 183,474 AI-related cases have been filed in China, a significant jump from last year.
More risks are on the horizon: your face could be cloned by AI, used to hawk products in short videos or star in web series without you even knowing. Your AI agent might sign a contract on your behalf, and if something goes wrong, no one can figure out who’s to blame.
AI is blurring the line between humans and tools. Where should the law draw the line? At WAIC 2026,Qiao Baojie, Dean of the Law School at Shanghai University of International Business and Economics,sat down with us to break down these issues from a legal lens. Hislaw school,built on a foundation in international economic law, has deep roots in foreign-related legal matters and runs the Shanghai Foreign-Related Legal Research Base. The “World AI Rule of Law Blue Book” is now compiled right here.
The Age of Agents: Who Deploys, Who Benefits, Who Bears Responsibility
“Cases of AI deepfake digital infringement and personal data protection are already piling up in courts,” Qiao says. Courts have started hearing these cases and have built a solid set of reference rulings.
Legally, AI deepfakes are just a new form of infringement, violating personal rights or privacy. Qiao believes the old legal rules aren’t useless—they just need to be rethought for new scenarios.
Another hot topic at this year’s WAIC: the legal headaches from AI agents. These agents are everywhere, helping folks plan trips, book flights, sign contracts, and even complete deals on their own. AI’s decision-making skills are getting scary good.
But here’s the rub: if an agent signs a deal that goes south, who’s on the hook? The user? The developer? The agent itself?
“AI is definitely a tool,” Qiao stresses. That’s the basic legal bottom line. So, the principle for responsibility boils down to nine words: who deploys, who benefits, who bears responsibility. If a user gives a command that sparks a dispute, they’re on the hook. But if the model has a security flaw or the operator didn’t check compliance, then the developer or operator takes the blame.
The Blue Book features a classic case from Hangzhou. After drinking, Wang Moumou fired up his AI-assisted driving mode,and the Hangzhou Linping District People’s Court found him guilty of dangerous driving, sentencing him to one and a half months in detention plus a 4,000 yuan fine. This ruling drew a clear line for “human-machine responsibility” in assisted driving: when the tech (like L2-level assistance) can’t handle all scenarios, human drivers can’t just hand over their duty to monitor.
But real-life problems are messy. Agents’ decision ranges, the boundaries of their authority, and tracking responsibility across different scenarios—there’s no unified written rulebook yet. Qiao admits this is the biggest legal black hole in the agent world right now.
China’s Got the Confidence for AI Governance
Zooming out from specific cases, the global AI governance landscape is shifting.
Right now, two main logics are emerging: one from the West (US and Europe) that prioritizes “individual rights + tech-centric” values. The other, from China, focuses on “human well-being + balancing development and safety.”
The WAIC 2026 chair’s statement highlighted a “global solidarity” spirit, and the release of the “Global AI Governance Initiative” shows China is stepping up to shape the global AI conversation. AI governance has hit a whole new level.
“China’s approach is shaking up the old Western-dominated model, where powerful countries set the rules and weaker ones just have to opt in or out—a kind of hegemonic, winner-takes-all thinking,” Qiao tells us.
He’s confident that as China gains more AI governance experience, its “circle of friends” will grow. “We’re developing AI fast, facing more problems, and we’ve got to solve them just as quickly. Other countries haven’t hit these issues yet, but they can learn from our path down the road.”
China has a full AI industry chain, a systematic legal practice, and long-term international academic platforms—plenty of confidence to weigh in on global AI rule-making.
The “World AI Rule of Law Blue Book,” compiled by the Shanghai University of International Business and Economics, debuted at WAIC in 2019 and has been published for eight straight years. It’s known in the industry as the “bible” of AI law. The English version is distributed globally by Springer, and starting in 2026, it’ll also be a CNKI journal.
“The Blue Book aims to solve two core issues,” Qiao says. “First, fill the gaps in global AI legal research by building an authoritative database of cases and theories. Second, help developing countries build their own AI governance capacity.”
For developing nations in the Global South, this is a ready-made “homework” to copy from. The Blue Book offers a governance path that’s different from the single Western standard. It means they don’t have to start from scratch on AI governance, slashing the trial-and-error costs of building new rules.
Law Can’t Keep Up with Tech, But It Can Set the Bottom Line
There’s a nagging tension between tech and law: AI upgrades every month or even week, but laws take years just to draft and pass.
“Law is about adjusting production relations. It only comes into play when those relations have evolved enough to need it,”Qiao explains. Law is naturally slow, but its core job is to “set the bottom line.” He introduces a concept: agile governance. Outside that bottom line, the law needs to react fast. How? Through “regulations + standards + filing and evaluation + sandbox trials.” That’s the path China is on.
Today, China has built an eight-tier progressive AI legal framework, from the Constitution to international governance cooperation. From the “New Generation AI Development Plan” to the “Interim Measures for Managing Generative AI Services,” from the “Algorithm Recommendation Management Regulations” to local laws, the structure is getting clearer.
Still, there are plenty of legal gaps to fill. For instance, there are no clear rules on AI-generated content copyrights. Open-source model licenses and infringement standards are fuzzy. The AI liability system isn’t fully fleshed out. And laws on data elements and computing power trading are lacking.
At the WAIC venue, we saw students from various universities zipping through forums and exhibit halls, getting a firsthand look at the tech changes. The law school at Shanghai University of International Business and Economics sends its students en masse to intern at courts, prosecutors’ offices, and law firms every year, pushing for a blend of education and industry.
Qiao believes that in the AI era, legal education needs to be redefined in two directions: foreign-related law and digital law. From the basics of AI tech to cross-border data compliance, from intellectual property to anti-sanctions, students have to learn a lot more than the traditional law curriculum offers.
