With DeepSeek’s AI breakthrough sending US tech stocks tumbling, a conversation has arisen regarding the juxtaposition between IP strategies in the East versus the West.

OpenAI cries foul, claiming IP theft based on US standards of software copyright and trade secret protection. DeepSeek, on the other hand, operating within China’s evolving IP landscape, would likely argue its actions fall within the bounds of ‘fair use’ or permissible data scraping under Chinese law.

This difference in legal interpretation is fuelling a transpacific IP tug-of-war, which will only intensify as new Chinese AI companies and their US-based competitors continue to jostle for prime position. So, how can the industry look to move forward together in greater harmony?

How US-China IP frameworks impact the AI race

The US and China approach IP protection from significantly different perspectives, creating a fundamental divide in their respective legal frameworks.

The US traditionally favours a ‘light-touch’ approach to regulation, prioritising innovation and market growth. This often translates to a reliance on self-regulation by the industry, with voluntary guidelines and frameworks playing a central role.

While agencies like the Federal Trade Commission (FTC) actively police deceptive and unfair practices related to AI as they did with five AI firms last year, the overall regulatory environment remains relatively loose, allowing for rapid development and deployment of AI technologies.

In contrast, China adopts a more proactive and interventionist approach to AI regulation, emphasising state oversight and guidance. The government views AI as a strategic technology crucial to achieving economic goals and has been actively promoting its development and adoption. This proactive stance is reflected in China’s evolving IP landscape, which includes specific regulations for generative AI, data scraping and other AI-related activities.

This framework, while still evolving, offers several potential benefits for innovators. For example, China’s legal system recognises the importance of agreements in determining copyright ownership of AI-generated works. This emphasis on official contracts can provide clarity and security for innovators, allowing them to define ownership and usage rights upfront.

Furthermore, China’s courts have been proactive in interpreting and applying copyright law to AI-generated outputs. Notably, cases like Yunkai Li vs. Liu (Springwind case 2023) and Cheng Lin vs. Hangzhou Gaosi, Changshu QinHong (Heart case 2024) have provided valuable insights into copyright protection for such works, offering guidance and precedents for innovators navigating this complex landscape.

This fundamental difference in approach to IP regulation on AI lies at the heart of the transpacific IP tug-of-war, making it challenging for new players from either side of the world attempting to break into new markets.

The challenge of cross-border IP enforcement

Cross-border IP enforcement in the context of AI presents a formidable challenge, exacerbated by the significant differences in legal frameworks between countries like the US and China.

These differences extend beyond just the interpretation of specific IP rights, encompassing broader issues of data privacy, competition law and even the fundamental philosophies underpinning IP protection.

One of the primary challenges lies in the varying standards for determining what constitutes infringement. For instance, the application of China’s Anti-Unfair Competition Law to data scraping activities involves a nuanced assessment of factors such as competitive relationships, the nature of the scraped data and the potential impact on market competition. 

Also, in a very recent case published in China titled Shanghai Character License Administrative vs. Hangzhou Small Design (Ultraman Hangzhou case 2024), the court expressed the view that a relatively lenient and inclusive standard should be adopted when assessing copyright infringement for data input and data training of AI models, which leads to eligible fair use defence. This approach contrasts with the stricter US stance, which often categorises data scraping as a violation of copyright or trade secrets laws.

What’s more, evidentiary issues complicate cross-border IP disputes. Gathering evidence of infringement in a different jurisdiction can be a logistical and legal minefield. Differences in discovery procedures, data privacy regulations and even language barriers can hinder the collection and admissibility of evidence.

Jurisdictional complexities also play a significant role. Determining which court has jurisdiction over a cross-border IP dispute can be contentious, particularly in cases involving online infringement where the infringing activity may occur in multiple jurisdictions.

This lack of harmonisation creates uncertainty and inconsistency in cross-border enforcement, begging the question: can a unified approach to AI-related IP emerge at a global level?

International AI agreements or industry standards

Given that the US and UK failed to sign the Paris Summit’s inclusive declaration on AI, it’s clear that it will take some time before we see any form of true international standards.

However, to address the challenge of IP enforcement and foster a more harmonious global AI industry, international agreements and the development of common standards for AI-related IP are essential.

These initiatives could include harmonising definitions of key terms like ‘fair use’ and ‘data scraping’; establishing clear guidelines and processes for cross-border IP enforcement; creating mechanisms for dispute resolution; and agreeing on ethical norms, safety measures and transparency guidelines for advanced AI technologies through bilateral and multilateral AI governance.

Industry-led best practices and standards could also play a significant role in bridging the gap between differing legal frameworks. Collaborative initiatives between US and Chinese AI companies could help establish common ground on IP protection, fostering a more cooperative and innovative environment to the benefit of all.

The transpacific IP tug-of-war in the AI industry requires a concerted effort from governments, businesses, legal players and international organisations which will evidently be no mean feat. However, by putting in the hard work to develop clear international standards and frameworks, we can unlock the full potential of AI for all while ensuring fair and effective IP protection across borders.

Landy Jiang is managing partner at specialist Chinese IP law firm Lusheng.