On 18-19 January 2026 the Central Political and Legal Work Conference was held in Beijing, setting out priorities for the year ahead. On two tracks — comprehensive cyberspace governance and technology development — the conference touched on artificial intelligence.

Front one: cracking down on AI-generated disinformation

In deploying comprehensive cyberspace governance, the conference called for advancing the “Qinglang” campaign, cracking down on online rumours, cyberbullying and other disorder, and on crimes that infringe personal information; it also stressed strengthening research on new technologies, guarding against the use of encryption such as blockchain to evade regulation, and cracking down on the use of AI to generate and publish disinformation.

This language brings AI-generated disinformation within the scope of the “Qinglang” campaign. For content platforms and AI service providers, it means content-review duties will extend from “post-hoc handling” towards “in-process interception”.

The conference called for accelerating the building of digital platforms for political and legal work, covering public security and stability, law enforcement and justice, team building and intelligent office work, under a model of “unified planning, two-tier construction, layered application”. It also asked political and legal units to advance their own technology programmes, grasp technology trends, deepen the integration of technological innovation with political and legal work, and actively explore AI applications.

This means the judicial and law-enforcement systems will become important AI application scenarios. For companies supplying AI products to this sector, data security, algorithm compliance and confidentiality will be core entry requirements alongside technical capability.

Front three: continued pressure on personal information protection

The conference placed “cracking down on crimes infringing personal information” alongside AI-generated disinformation, pointing to a single risk chain: AI lowers the cost of forgery and dissemination, while personal information is often the material that is forged and misused.

This is consistent with recent enforcement trends. See our notes on the CAC’s crackdown on AI content lacking labels and enforcement of synthetic-content labelling.

Four pointers for companies

First, content labelling is a baseline duty. Providers of generated synthetic content must ensure labelling obligations are met; see our note on the verification duty of generative AI users.

Second, face and voice features carry the highest risk. Features involving face generation or swapping and voice cloning should be assessed for authorisation and purpose boundaries; see the fifteenth batch of deep-synthesis filings and our note on face-recognition filing.

Third, build a reporting and handling loop. Platforms should provide convenient reporting channels and retain handling records; see reporting channels for AI misuse.

Fourth, political and legal projects need compliance designed in from the start. Companies participating in this digitalisation should prepare data-security and confidentiality plans at the bidding stage; see our data compliance service.

The conference’s language brings AI governance from “sector regulation” further into a “social governance” framework. For companies, the reference points are no longer only sector regulators, but also the judiciary’s adjudicative orientation and enforcement standards.

This article is based on public reporting of the 2026 Central Political and Legal Work Conference. It is for general reference only and does not constitute legal advice.