Recently, a Chinese AI startup named DeepSeek has garnered widespread attention globally. On January 27, 2025, DeepSeek's AI assistant simultaneously topped the free charts of the Apple App Store in both China and the United States, becoming the first Chinese tech product to achieve a double-chart victory. DeepSeek has received market attention for its high-performance, low-cost AI models. The DeepSeek-R1, for example, has a training cost only onethirtieth of that of similar models from OpenAI, and it has demonstrated strong capabilities in areas such as natural language processing and computer vision.
The emergence of DeepSeek has disrupted the market landscape dominated by traditional AI giants. Its open-source and low-cost approach has posed a significant challenge to hardware suppliers like Nvidia and software giants like OpenAI. By innovating algorithms, DeepSeek has reduced the reliance on high-performance chips, thereby challenging the U.S. monopoly on computing power. The lower development cost of its models has also shown new possibilities for developing high-performance AI models.
However, this has also led to a series of controversies regarding intellectual property, trademarks, and data privacy, which serves as a warning bell for Chinese tech companies looking to expand overseas.
1. An Overview of Legal Disputes DeepSeek is Facing
- Intellectual Property Claim
OpenAI claims that DeepSeek used its proprietary models and techniques to develop its own AI system. Specifically, OpenAI alleges that DeepSeek employed a technique called "model distillation," which involves training smaller AI models using outputs from larger, more advanced models.
OpenAI's bases its claims on the following grounds: (a) DeepSeek allegedly violated OpenAI's terms of service by using its models to create a competing AI system; and (b) OpenAI argues that DeepSeek's actions constitute a breach of intellectual property rights, as they used OpenAI's proprietary technology without permission.
OpenAI claims that it has detected evidence that DeepSeek engaged in "distillation" by extracting large amounts of data through OpenAI's API. This evidence was reportedly discovered by Microsoft security researchers. OpenAI has presented several pieces of evidence to support its claims, including (a) a finding made by a security researchers at Microsoft that individuals linked to DeepSeek were withdrawing large amounts of data through OpenAI's API; (b) performance of DeepSeek's AI model, R1, that is similarly to OpenAI's models but at a fraction of the cost and with less computing power, suggesting the use of OpenAI's outputs; and (c) instances where DeepSeek's model reportedly referred to itself as being developed by OpenAI.
DeepSeek has not publicly responded to OpenAI's claims.
- Trademark Issue
DeepSeek filed an application with the U.S. Patent and Trademark Office (USPTO) to trademark its AI chatbot apps, products, and tools. However, they were beaten to the punch by just 36 hours by Delson Group Inc., a Delaware-based company. Delson Group claims it has been selling AI products under the "DeepSeek" brand since early 2020.
Delson Group asserts that it has prior rights to the "DeepSeek" trademark. The company's CEO, Willie Lu, has a history of trademark squatting—registering trademarks with the intention of selling them later or capitalizing on a brand's popularity.
- Data Privacy Concerns
Data is the lifeblood of artificial intelligence. It fuels the training of AI models, enabling them to learn, adapt, and improve their performance. Without diverse and extensive datasets, AI systems would struggle to understand and respond accurately to real-world scenarios. Moreover, AI operates by collecting and interacting with users' data, too.
As soon as DeepSeek topped the free charts of the Apple App Store, it attacks attentions from various data privacy enforcement bodies. The Italian data protection authority (GPDP) has demanded information from DeepSeek about its handling of user data, including the type of data collected, how it is stored and whether that data is stored on Chinese servers. It is out to ensure compliance with the EU's GDPR rules. The Indian government stated that DeepSeek needs to be tested on security parameters will be tested and it also said “that DeepSeek is an open-source model, and we are very soon going to host DeepSeek on Indian servers, the way we have hosted Llama on India servers.
2. A Brief Discussion of Legal Issues Surround Deep Seek
It appears that DeepSeek is facing a series of legal challenges when it attracts attention globally. Some of these issues are new and some are old-fashioned.
- A Discussion about Intellectual Property
Claim
Generally, proprietary models and techniques are those that are protected either as patents or trade secrets. Since OpenAI has not claimed that it has patented the “distillation model” which it asserted DeepSeek has infringed against, it is reasonably to presume that it is protected as a “trade secret.” However, when a model is openly available for use, it typically means that it is accessible to the public without restrictions, and anyone can use, modify, or build upon it. However, the situation becomes complex when an openly available model is used to create a new model that competes directly with the original. Does this an unfair competition practice?
In the case of DeepSeek, OpenAI alleges that DeepSeek used outputs from OpenAI's models to train its own competing AI system. This raises questions about whether the techniques used to create the new model can still be considered proprietary if they are based on outputs from an openly available model. This is not a simple question, especially because the trade secret rules in US are state laws, which are common law; the ruling on this issue will based on precedents, equity principles, and a balance between encouragement on innovations and public interest.
OpenAI also claims that its terms of service explicitly prohibit using its services or outputs for any commercial purposes that compete with OpenAI and that DeepSeek violates these terms, resulting in intellectual property infringement. Question remains whether the unilateral terms of service could create right in rem.
- A Discussion about Trademark Issue
Although Delson Group applied for the DeepSeek trademark earlier, DeepSeek can still get the trademark back under according to the first-to-use doctrine.
The U.S. follows the first-to-use trademark system, under which marks that are actually used in commerce first are assigned trademark rights and given priority in their territory of use. Under firstto-use trademark regimes, being the first to apply for trademark registration does not guarantee priority to the applicant. As such, if DeepSeek can prove that it was the first to use the "DeepSeek" trademark in commerce, it may have grounds to challenge Delson Group's registration.
Bad faith filing is also a ground which may help DeepSeek to get its trademark back. Under U.S. laws, a bad faith claim can be raised if there is evidence that the trademark registration was filed with the intent to deceive or harm the rightful owner. The Lanham Act, which governs trademarks in the U.S., allows for the cancellation of a trademark registration if it was obtained in bad faith. Typical indications of bad faith filings include a history of trademark disputes or an apparent lack of genuine interest in the mark. For instance, if an applicant files for a trademark primarily to block competitors, rather than to use the mark in commerce, this may signal bad faith. The fact that Delson Group filed a large number of famous trademarks before could be an evidence.
- A Discussion on Privacy Concern
Obviously, foreign governments have imbalanced concerns about AI operated by Chinese companies. They are concerned that China government may have access to data stored in China and could request data localization. As NBC commented “Under Chinese law, all companies must cooperate with and assist with Chinese intelligence efforts, potentially exposing data held by Chinese companies to Chinese government surveillance. That system differs from the U.S., where, in most cases, American agencies usually need a court order or warrant to access information held by American tech companies.” This opinion may not be true as US federal government has extensive authority on compelled disclosure of data on CLOUD Act but it reflects the universal concerns that the PRC government could use this access for espionage, economic advantage, or other purposes.
Additionally, when Chinese AI systems are going internationally, they encounter different regulatory regimes with varying requirements on using anonymized data. Different countries may have specific requirements for anonymizing data to ensure privacy and compliance with local data protection laws. Navigating these regulatory landscapes can be challenging for AI companies, as they must ensure compliance with diverse and sometimes conflicting requirements. This can impact the design, development, and deployment of AI systems across different regions.
3. Suggestions to Chinese Companies Going Abroad
Expanding business abroad is challenging, not only from a business aspect, but also from a legal aspect. It requires that Chinese companies have a prospective view on how and where their business will expand and consider the legal environment in the targeted jurisdictions.
To clear intellectual property related risks, it is advisable for Chinese companies to deploy their intellectual property rights in target jurisdictions before launching its business there. Otherwise, trademark squatters or other IP “occupations” could be troublesome. Moreover, it would also be advisable to perform a freedom-to-operate analysis and trademark availability search in advance to clear IP risks at a lower cost.
Data privacy is an emerging legal area, especially in the AI industry, where the laws in many jurisdictions are developing. Chinese AI operators need to consider regulatory requirements in various target countries and may potentially build their products reflecting these requirements.
Note, importantly, laws in foreign jurisdictions could be quite different. A comprehensive understanding about the laws and legal cultures in major target jurisdictions could be important for expanding business abroad.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
