The Cyberspace Administration’s annual ‘Qinglang’ campaign arrives in a materially different regulatory environment to last year’s edition, and in the same week the White House accused China of running ‘industrial-scale’ AI theft operations.

China has launched a months-long enforcement campaign targeting the misuse of artificial intelligence, according to Reuters.

The campaign, initiated by the Cyberspace Administration of China (CAC) and coordinated with the Ministry of Public Security and other agencies, targets AI-enabled fraud, deepfakes, disinformation, and illegal applications that violate privacy and intellectual property rights.

The action is the 2026 edition of what has become an annual enforcement mechanism, the ‘Qinglang’ (Clear and Bright) special campaign series. Its immediate predecessor, launched on 30 April 2025 and titled ‘Rectification of AI Technology Misuse,’ ran for three months across two phases.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

By the time its first phase concluded in June 2025, authorities had taken down more than 3,500 AI-related products, scrubbed over 960,000 pieces of illegal or harmful content, and shut down or penalised more than 3,700 accounts.

This year’s campaign arrives in a substantially more developed regulatory environment and against a geopolitically charged backdrop that makes its scope and targets distinctly more complex than its predecessor.

What the campaign targets?

China’s AI abuse enforcement campaigns are structured around a taxonomy of misuse that has expanded with each iteration as both the capabilities and the criminal applications of AI have advanced.

Based on the established Qinglang enforcement framework and the new regulatory measures enacted in 2025 and early 2026, this year’s campaign is expected to target several categories simultaneously.

The first and most commercially significant is AI-enabled fraud and impersonation. China has seen a dramatic increase in the use of voice-cloning and face-swapping deepfake technology to impersonate celebrities, business executives, and government officials in scams targeting ordinary consumers.

The CAC’s 2025 campaign specifically targeted the use of AI to ‘impersonate relatives and friends and engage in illegal activities such as online fraud’ and the ‘improper use of AI to resurrect the dead’, a reference to the use of AI-generated likenesses of deceased people without consent.

The CAC published draft rules for digital virtual human services on 3 April 2026, covering consent requirements for likeness use and banning the bypass of biometric authentication systems, with a public comment window that closed on 6 May.

The second major category is AI-generated disinformation and ‘online water army’ activity, the industrial-scale use of AI to create fake social media accounts, generate and distribute coordinated content, manipulate engagement metrics, and create artificial trending topics.

The 2025 campaign identified this as a priority for its second phase, focusing on platforms facilitating AI-powered account farming, batch content generation, and social bot networks.

Third is non-compliance with mandatory filing and registration procedures. China requires that large language models offering generative AI services to the public undergo a security assessment and complete a filing with the CAC before launch.

As of March 2025, 346 gen AI services had completed the LLM filing; many more had not. The 2025 campaign’s first phase identified unfiled AI products as a primary rectification target, with local regulators including the Shanghai CAC penalising three AI applications that had provided services without completing the required process, and the Zhejiang CAC ordering app stores to remove a face-swapping app that had not undergone security assessment.

Fourth is the management of training data, specifically, using training corpora that incorporate content infringing on intellectual property rights, privacy rights, or consent obligations.

This enforcement angle is particularly sensitive in 2026 given the White House’s formal accusation, issued on 23 April, that Chinese companies are running ‘industrial-scale’ distillation campaigns to extract capabilities from US frontier AI models using jailbreaking techniques and tens of thousands of proxy accounts.

China’s domestic enforcement campaign does not address that US accusation directly; it is focused on protecting Chinese rights holders and users, not American ones. But the two regulatory environments are now evolving in explicit awareness of each other.

The 2026 campaign operates against a substantially more developed domestic regulatory architecture than its predecessor. Several major rules either took effect or were published in draft form in the months leading up to this enforcement push.

China’s mandatory AIGC (AI-generated content) labelling standards, requiring visible and technical labels on all AI-generated text, images, audio, and video, took effect on 1 September 2025.

On 10 April 2026, the CAC published the Interim Measures for the Management of Anthropomorphic AI Interactive Services, governing chatbots, AI companions, and AI customer service agents that simulate human personality and communication styles, with effect from 15 July 2026.

On 3 April, the CAC published draft rules for digital virtual human services covering biometric deepfakes, with public comment closing on 6 May 2026. And in April 2026, the CAC, MIIT, and MPS jointly published a 2026 personal information protection enforcement agenda targeting seven sectors including internet advertising, education, healthcare, and criminal data trafficking.

The effect of this layered rulemaking is that the 2026 Qinglang campaign has substantially more legal teeth than its 2025 predecessor. Enforcement actions can now be grounded in the mandatory labelling standards, the LLM filing requirements, the Deep Synthesis Measures from 2023, and the new Anthropomorphic AI Interim Measures, all simultaneously.

Companies found to be in violation face administrative penalties, service suspension, mandatory rectification within fixed deadlines, and in serious cases, criminal referral to public security authorities.

The geopolitical framing

The campaign’s timing gives it an unavoidable geopolitical dimension. One week before its launch, on 23 April, White House OSTP Director Michael Kratsios published a memo accusing China of running ‘deliberate, industrial-scale campaigns to distil US frontier AI systems,’ using tens of thousands of proxy accounts and jailbreaking techniques.

The accusation was framed as evidence that Chinese companies, including DeepSeek specifically, were stealing US AI intellectual property to train near-frontier models at a fraction of the cost, in violation of US labs’ terms of service.

China’s domestic AI misuse campaign is not a response to that accusation, it is an annual structural exercise that predates the current geopolitical moment. But its focus on AI-generated fraud, impersonation, and data rights violations does implicitly reinforce the argument that both countries are grappling with the same AI misuse categories from opposite ends: China policing its own population’s exposure to AI-enabled fraud and manipulation, while the US accuses Chinese actors of using AI to defraud its own technology companies of proprietary capabilities.

The campaign also runs alongside the Trump-Xi summit scheduled for 14 May in Beijing. The White House memo on AI theft was explicitly noted by analysts as potentially complicating those talks, with AI and semiconductor export controls both expected to be on the agenda.

China’s domestic enforcement campaign, framed around protecting users and maintaining social stability, is unlikely to feature in those diplomatic discussions directly, but it forms part of the broader picture of how each government is positioning its domestic AI governance in the lead-up to a high-stakes bilateral meeting.