As Chinese companies graduate from contract manufacturing to proprietary brands and technology exports, the intellectual property rules originally written for the ‘world’s factory’ are undergoing an upgrade. Pan Xinyi reports
At a gas station in California, a knockoff figurine nicknamed “Lafufu” by netizens sits by the checkout, priced at USD4.99. It shares a face with Labubu, a viral designer toy from Chinese brand Pop Mark, and in all likelihood may share a production line in Dongguan, China. Counterfeiting long ago ceased to be a local night-market tale; it has become a structural byproduct of China’s supply-chain efficiency, leaking silently across the globe.
This is no isolated case. When Chinese factories begin to build their own brands and technologies, the intellectual property (IP) map drawn for the era of original equipment manufacturing no longer charts the waters they sail today.
What China is undergoing is not a moral awakening, from infringer to rights holder, but a system-wide upgrade driven from within. As industry climbs the value chain, as AI front-loads risk into the very first line of code, as companies venturing abroad repeatedly hit the high, well-guarded patent walls of mature markets, the domestic institutional foundations underpinning this contest still bear the casting marks of two decades ago.
The core IP framework of China took shape around the country’s accession to the World Trade Organisation, designed to serve an export-led economy, secure trade access, and attract foreign investment. Its centre of gravity was rapid rights registration, not innovation incentives.
Today, China has moved from contract manufacturing to home-grown innovation. According to WIPO’s Global Innovation Index 2025, it has for the first time broken into the top 10, the only middle-income economy in the top 30. Yet the framework remains largely unchanged.
Pressure is showing up everywhere. China’s National Intellectual Property Administration reports that trademark applications exceeded 6 million, with examination cycles compressed to four months. However, Marks & Clerk, a UK IP firm, notes that to keep up the speed, the examination authorities have leaned heavily on assistant examiners, and inconsistencies in standards and rising refusal rates have followed.
The judicial side is under equal strain. In 2025, the Supreme People’s Court’s Intellectual Property Tribunal awarded punitive damages in 30 cases, totalling about RMB1.13 billion (USD165.6 million), more than half the cumulative sum in the tribunal’s seven-year existence.
Yet China’s punitive damages regime has only been rolled out gradually since 2019, and remains in a phase of adaptation and refinement; calculating damages is complex, and the criteria for application are still not fully clear. At a time when judicial protection has greatly intensified, adjudication is only growing harder.
The IP system built for the world’s factory is being stretched out of shape by its very users. From trademarks to trade secrets to patent commercialisation, the mismatch between old maps and new terrain points to a single question: now that IP has evolved from a passive compliance accessory into infrastructure for global competition, should companies continue to follow the existing chart, or draw their own course?
Trademark: use it or lose it
Close to 50 million active trademarks sit on China’s register, a substantial share of which have never been genuinely used. This is not a hallmark of market vitality, but the institutional reckoning for decades of prioritising registration over use. The current revision of the Trademark Law takes direct aim at this deep-rooted malady.
Ma Dongxiao, a Beijing-based partner at Zhong Lun Law Firm, distils the core logic of the revision as a systemic shift from “registration first” to “use first”. He notes that for the draft revision, three fronts advance in parallel, each targeting a stubborn malady: hoarding; drawn-out timelines; and irregular use.
On the registration side, a dedicated new chapter requires a return to the original purpose of use and curbs hoarding that exceeds normal business needs. On the procedural side, shortened opposition periods and a new “suspension of examination” mechanism address the wheel-spinning caused by changed circumstances. On the administrative side, compliance obligations on trademark agencies are simultaneously tightened.
For businesses, the most immediate shock falls on existing portfolios. Ma believes that stockpiled defensive trademarks will face a dual test of “intent to use” and “normal business need”, and that companies must “complete the transition, in both mindset and behaviour, from registration management to use management, so as to adapt to the new trademark regime as quickly as possible”.
The central question of trademark stewardship is thus shifting from securing rights to building a watertight chain of evidence.
The institutional correction has another target squarely in its sights: the shadow industry of profiteering through registration and litigation. Zhang Jian, a Beijing-based partner at Han Kun Law Offices, points out that the draft revision upgrades the regulation of bad-faith squatting for the first time, moving beyond a single track of “administrative rejection or invalidation” to a dual-track remedy of “administrative fines plus civil damages”.
Victims of squatting can now file civil suits for compensation directly, transforming the economic calculus from near-costless rejection to the prospect of fines and damages combined.
Beyond squatting, abusive litigation is also in the crosshairs. The draft introduces a counter-compensation clause, making explicit that parties who maliciously initiate trademark litigation causing losses to others shall bear civil liability, a provision that takes direct aim at shakedown-style lawsuits brought purely for profit.
Beyond remedying old ills, the regime is also clearing a path for new realities. Hatty Cui, the general manager of China at Rouse in Beijing, notes that the draft proposes to bring dynamic marks into the fold of registrable trademark types, responding to the evolution of brand expression in the age of short video and AI.
She advocates for a design philosophy of “broad admission, strict delimitation, allowing more new types of marks to gain registration at the gateway, while applying rigorous scrutiny to the boundaries of their rights”. The approach is pragmatic but far from straightforward. Assessing the distinctiveness and stable presentation of dynamic marks is considerably more complex than for conventional trademarks, and the difficulty of establishing and proving rights climbs in tandem.
Another proposed amendment has sparked wider debate. The “not for the purpose of use” clause now drops the “bad faith” qualifier, casting the regulatory net appreciably wider. Cui cautions that, without clear criteria for differentiation, a net wide enough to catch hoarding applications may also snare legitimate defensive filings. The regime, she argues, must “strike a balance between ‘emphasising the use attribute’ and ‘respecting normal business arrangements’”.
The direction of the trademark cleanup is clear, but where exactly the line falls between hoarding and legitimate reserves matters just as much. Cui suggests that “for registered trademarks that have been used continuously over a long period, and have established a stable market order, due respect should be given to the goodwill accumulated through bona fide use and the registrant’s legitimate reliance interests”.
Even as the regime accelerates its course correction, it must preserve a sense of security for those who have played by the rules. The leap from a registration mindset to a use mindset is ultimately a test not just of force, but of finesse.
When trade secrets walk
In the real world of commercial competition, the ways a company loses its edge are often more insidious, and more lethal, than having a trademark squatted. What keeps businesses awake at night is rarely a single dramatic heist of an entire technology stack. It is the drip-feed of attrition: a key employee defects, data is copied, algorithms are migrated, distribution channels are reassembled. Competitive advantage is dismantled piece by piece in transit and reconstituted in a new setting.
Tang Qinglin, founding partner of Yunting Law Firm in Beijing, observes that in the past year trade secret infringement cases involving technical know-how have clustered heavily in new energy, semiconductors, AI, biopharmaceuticals and advanced manufacturing.
The scenarios follow a familiar script. A lead battery engineer at a new-energy firm jumps ship carrying cell formulations, and within six months a rival launches a product with near-identical performance specs; an AI company’s head of algorithms leaves to start his own venture, and the new firm’s model architecture is virtually indistinguishable from his former employer’s technical roadmap.
The evolution of rules governing unfair competition and trade secrets is a direct response to this reality. Tang notes that judicial reasoning has undergone a systemic shift.
On the evidentiary front, rights holders need only make a prima facie showing of secrecy, protective measures, access and substantial similarity before the burden shifts to the defendant. On the substantive front, courts increasingly assess a technical solution as an integrated whole; partial disclosure of individual elements no longer readily defeats a claim of overall secrecy.
At the same time, punitive damages are becoming routine. Judgments applying the statutory ceiling of one to five times actual damages appear with growing frequency, rewriting the old calculus in which the cost of infringement ran cheaper than the cost of enforcement.
On the execution front, the criminal-civil interface has tightened markedly. Carrying technical information out in one’s head, so-called “memorised misappropriation”, and electronic intrusion have both been brought within the scope of criminal liability, ratcheting accountability well beyond civil compensation.
Notably, a 2020 Supreme People’s Court judicial interpretation on trade secret cases explicitly lists data, algorithms and computer programs as protectable “technical information” under the Anti-Unfair Competition Law, giving companies asserting confidentiality over digital assets a clear legal footing for the first time.
Yet the most distinctly contemporary challenge may be that “secrecy” itself is becoming harder to define. Zhang, of Han Kun, says: “The crown jewel assets of an AI company, model architecture, weight parameters, training data, are highly non-intuitive and highly portable.” How to bring these intangible technical assets within a trade secret management framework, and how to delineate their precise “secret points” in litigation, is as much a conundrum for courts as it is the hardest blind spot for companies to police in day-to-day operations.
The other side of the predicament is that companies themselves are woefully under-prepared. Eva Tong, a general counsel at Li Ning, identifies trade secrets as the single area of greatest concentrated compliance risk for businesses today. She observes that the leakage threat wears a different face depending on the role: salespeople walk away with client lists; R&D staff with test data and process parameters; senior executives with strategic plans and undisclosed financials, yet most companies’ confidentiality defences remain one-size-fits-all and vague.
Richard Tian, legal director of Skechers China, corroborates the diagnosis. At many companies, non-disclosure agreements have never classified confidential information by tier, and system access is not revoked promptly when employees leave. The gaps are rarely in any single link; rather, the entire management chain lacks institutional scaffolding.
He says that the risk does not flow in one direction only. A company can be the victim of trade secret theft and, in the same breath, an unwitting infringer of someone else’s IP. This dual exposure points to a more fundamental problem: most companies’ IP management remains fragmented and reactive, with the absence of an end-to-end IP governance system spanning identification, classification, protection and accountability.
In certain industries, the fragility of trade secrets stems not just from management gaps but from the physical nature of the asset itself. Leon Li, a legal manager, APAC, at Driscoll’s Management, faces a predicament with an especially punishing time structure: berry breeding cycles (developing new varieties through cross-pollination) are measured in years yet the output, a single seed, is by its very nature a replicable, portable vessel of proprietary technology.
The gulf between the time invested in creation and the ease of duplication is immense. Core germplasm and breeding data can slip away the moment a researcher moves on, and at the retail end, look-alike packaging paired with a regional-origin endorsement is enough to sow confusion, leaving breeders struggling to convert their first-mover investment into brand recognition at the point of sale.
From battery formulations and model architectures to a single berry seed, the universe of assets that trade secret law must protect is expanding at speed, and the regulatory framework is racing to keep pace. The Provisions on Protection of Trade Secrets, set to take effect in June 2026, represent the first comprehensive regulation update in the field in three decades, codifying into a formal rule-set the protective logic that courts have been building case by case.
However, legal protection alone does not create value. A patent may be granted, a trade secret confirmed, yet between that recognition and its translation into products, revenue and competitive position, a long road remains.
Protecting data assets starts with R&D
AI training data is drawn from copyright regimes spanning multiple countries, and the moment a model goes live it touches users across jurisdictions.
Compliance pressure begins with the very first line of training code. China is now the world’s largest holder of AI-related patents, accounting for 60% of the global total, according to the World Intellectual Property Organisation.
China’s regulators have moved quickly. From the Provisions on the Management of Algorithmic Recommendations for Internet Information Services to the Interim Measures for the Management of Generative AI Services, Beijing has rolled out rules targeting specific AI use cases at a pace few jurisdictions can match.
A unified foundational AI statute, however, remains only a preliminary deliberation item on the NPC Standing Committee’s 2025 agenda, not yet formally in the legislative pipeline.
While the legislative architecture is still taking shape, the existing IP system is already struggling to keep data assets safe.
Lu Bin, general counsel and chief compliance officer at Inspur Digital, says that data readily loses its confidential character once it circulates inside an organisation, or is fed through model training, eroding the very premise on which trade secret protection rests.
Raw data, meanwhile, seldom clears the originality bar that copyright demands. In an AI context the two principal lines of defence can collapse at the same time, leaving core data assets in a protection vacuum.
Shielding one’s own data is difficult enough, but a company that also develops AI products may find itself on the other side of the line, encroaching on the data rights of others.
Zhang Jian, a Beijing-based partner at Han Kun Law Offices, traces three pressure points along an AI product’s life. The first sits at the front door: copyrighted images; articles; and personally identifiable data scraped by engineers to feed training sets. Every unlicensed item is a lawsuit waiting to happen, and the bigger the product’s commercial footprint, the larger the target on its back.
The last sits at the output end: when code or images generated by artificial intelligence generated content (AIGC) closely resemble existing protected works, liability falls not on the model but on the company that deployed it.
The risk most easily overlooked, though, lurks in the mundane space between the above-mentioned two. An engineer pastes proprietary source code into an external large model’s chat window for a quick debugging fix; an unreleased business plan is fed to an AI for a summary – and confidential information slips quietly out through the cracks in everyday workflows.
Whether they are guarding their own data or policing infringement risk in product development, companies confront the same time lag: technology moves daily, while the legal framework is still playing catch-up.
Lu insists the response cannot wait for legislation to arrive fully formed. “The in-house counsel’s function should be more than a risk controller, it should be a guardian of technological value,” he says.
From auditing the authorisation chain behind training data, to pinning down liability for generated content, to maintaining a living loop of algorithm filing and content safety review, compliance must be woven into the R&D process from start to finish.
Patent, and then what?
China does not lack patents. By the end of 2024, the country held more than 4.7 million valid invention patents, topping the global league table for several consecutive years. But numerical supremacy has not automatically translated into industrial advantage.
A vast stock of patents remains stranded in certificates and spreadsheets, unincorporated into standards, unembedded in products, unformed into tradeable capability. The next chapter for China’s patent regime is not about issuing still more certificates; it is about whether anything comes of the ones already issued.
Zhang, of Han Kun, notes that a great many patents owe their existence not to the pull of market demand but to the push of “qualifying for high-tech enterprise status or harvesting subsidies”. When the motive for filing is decoupled from commercial logic, a patent lacks a market foundation from the moment it is conceived. She also sees corporate attention now migrating, away from chasing volume and toward screening and maintaining high-value patents, earlier-stage global portfolio positioning, and staking out ground in standard-essential patents.
Li, of Driscoll’s Management, feels the problem from the other end. Working in the berry industry, he finds that “the real difficulty with patents is not deciding what to file, but the ‘last mile’, turning a granted technology into something that actually works”. Once a patent is in hand, the true sticking point is fitting it to supply-chain realities, aligning it with industry standards, and plugging it into a viable commercial scenario.
To close that gap, his team assembles cross-functional groups at the project initiation stage for each new variety, sorts patents into three operational tiers – “core defence” “collaborative opening”, and “standards participation” – and writes patent conversion rates into the KPIs of both R&D and business teams, so that every patent must answer, before it is even filed: Who does it protect? And who does it push back against?
Zhang Zhanjiang, a Beijing-based partner at Jingtian & Gongcheng, prefers to read the problem across the full supply chain: “The bottleneck in converting innovation into commercial reality is not a single point of failure, but a structural condition running through the entire pipeline, from supply to matchmaking to conversion to commercialisation.”
He notes that upstream, a mass of research output stalls at the laboratory stage, separated from industrialisation by what he calls a “technology-readiness chasm”. Midstream, newer mechanisms such as open licensing and use-now-pay-later arrangements appear to lower the threshold, yet in practice the chief beneficiaries remain large enterprises, the very small and medium-sized enterprises that most need the technology find it still out of reach.
Downstream, many companies continue to treat patents as achievements to display, rather than assets to put to work, with no closed loop connecting creation to monetisation.
In Zhang’s view, technical breakthroughs translate into lasting market value only when IP strategy is embedded across a company’s research, operations and competitive strategy.
Tech going global: first battle is not fought on shelves
The IP rules governing AI are still being written, but hard-tech companies heading overseas walk straight into fully mature international IP regimes. The more complete the legal toolkit, the more weapons rivals can reach for.
From customs clearance to retail shelves, every step can become a target for infringement claims, and litigation costs and market access risks pile up at every turn.
A single 12-month window tells the story. From late 2024 to early 2025 alone, Corning filed two back-to-back petitions with the US International Trade Commission (ITC) seeking section 337 investigations against Chinese display makers including Irico Display Devices, HKC, and TCL CSOT.
In July, South Korea’s Pantech lodged a further 337 case over Chinese mobile cellular devices, naming OnePlus, Lenovo and TCL among the respondents.
By November the ITC had launched yet another probe into certain liquid crystal display components, with HKC, TCL and Hisense on the respondent list.
Facing this barrage, Zhang Jian, a partner at Han Kun Law Offices in Beijing, suggests that companies erect compliance firewalls across their equity structures, data storage arrangements and IP-holding entities, ring-fencing IP risk so that a hefty sanction in any single jurisdiction cannot pierce through to the global parent.
Lily Li, a Palo Alto-based partner at Morrison Foerster, adds a different tactical lens drawn from the respondent’s side of the table. Citing her team’s defence of Insta360 against GoPro’s ITC complaint, she notes that the commission’s docket moves at breakneck speed.
“Once you’re inside the ITC process, success doesn’t hinge on fighting on every front, what matters most is pinpointing the decisive issue within a tight window and building an argument around it that can survive scrutiny,” she says. The case ended with a finding of no violation across every utility patent at issue.
As companies fight case by case on the frontline, the legal arsenal behind them is also growing. Article 33 of the newly revised Foreign Trade Law, which took effect on 1 March 2026, writes into national legislation for the first time a mandate to build an overseas IP early warning and rights protection assistance platform.
Article 35 goes further: Where another country or region fails to afford adequate IP protection to Chinese businesses, the competent authorities may take countermeasures, handing Beijing, for the first time, an explicit statutory basis for reciprocal action.
There is still daylight between ink on the statute book and enforcement on the ground but, at the very least, Chinese companies defending their rights abroad no longer do so with nothing at their backs.
Practitioner’s Perspective Article Series
Compensation not contingent upon actual sale in patent infringementBy Guo Xiaojun, CCPIT Patent & Trademark Law Office |
 |
Protection of non-traditional trademarks in ChinaBy Wang Xiao and Kelly Liu, Tahota Law Firm |
 |
