The rapid rise of artificial intelligence and large language models (LLMs) has given rise to a thorny legal dilemma: Who should own the rights to the news?

There are basically two separate but related issues at stake. First, AI companies like OpenAI and Meta, which rely on vast quantities of data to train their models, view news articles as sources of high-quality writing and information. But their harvesting and mining of this content has led to a spate of lawsuits from media outlets accusing them of using copyrighted content without compensating the original creators and publishers. Other news outlets, such as the Associated Press, have opted to collaborate with big tech, licensing their content to LLMs, with or without the consent of the original authors.

Meanwhile, newsrooms across the globe are turning to automated tools such as news bots to generate content. But the question of who owns the copyright for AI-generated articles remains a legal minefield — one that news organizations, already struggling with an economic crisis, collapsing audiences, and a decline in public trust, cannot afford to ignore.

Historically, news copyright has always been a delicate balancing act. On the one hand, the copyright system must protect the rights of journalists and publishers while incentivizing them to produce quality content. On the other, it also must guarantee the public’s right to be informed. As AI begins to play an expanding role in news production, this balance has only become more difficult to maintain.

Curiously, a Chinese court ventured into the murky territory of news copyright and AI all the way back in 2019, setting a precedent with far-reaching implications for the nation’s media.

The case revolved around Tencent’s Dreamwriter, an AI-powered news bot capable of churning out financial news updates in seconds. In 2019, Tencent sued another company for republishing a Dreamwriter-generated article without authorization. The Shenzhen Nanshan District Court sided with the tech giant, determining that, although the content was created by an AI system, there was significant human involvement in designing and operating this system.

Under Chinese law, this justified extending copyright protection to the resulting work.

Dreamwriter never lived up to the hype — its newsfeed shut down in July 2020, ending with an update on Chinese electric vehicle (EV) company Nio’s stock price. But the landmark ruling has only grown more relevant in recent years, as new, more advanced LLMs upend the media landscape.

Copyright protections for news articles in China are governed by the country’s Copyright Law. While “purely factual information” cannot be copyrighted, a 2021 revision to the law clarified that representations of facts — such as news articles that meet an originality threshold — are eligible for copyright protection. The clarification was aimed at supporting journalism in the face of algorithm-driven news aggregators developed by tech companies, many of which circulate content from news publishers without authorization or compensation.

However, the 2021 revision also extends protection to “other intellectual achievements conforming to the characteristics of works,” an umbrella category that strengthened the hand of tech firms by effectively broadening copyright protection for products involving innovative algorithms. The change essentially codified the Dreamwriter verdict, which also highlighted the intellectual efforts of software developers.

The copyright regimes of the European Union and the United States offer a useful comparison point. As in China, copyright rules in the two regions tend to prioritize corporate investment over individual agency or the public good. Indeed, although the EU is known for its rigorous regulation of AI and strong antitrust law, regulators there have been noticeably reticent to make copyright a choke point for tech companies.

This can be seen in how the EU’s Digital Services Act and Artificial Intelligence Act address AI-related challenges without centering copyright protections. The EU’s Copyright Directive also makes exempting works from text and data mining something rights-holders must opt out of, otherwise AI companies may use copyrighted materials for non-commercial research purposes without obtaining prior permission. While this exception is intended to support innovation, it also avoids creating overly restrictive conditions that could hinder the development of AI technologies.

Data mining has emerged as a particular point of contention in the U.S. For example, The New York Times has filed a lawsuit against OpenAI, accusing the company of using its copyrighted content without permission to train its AI models. Other major publishers have expressed concerns over the potential loss of control over their intellectual property as AI tools become more sophisticated.

Even companies that have opted to work with AI firms, like the Associated Press, have kept the exact details of their arrangements secret, leading to concerns about transparency and fairness in the negotiation process. While there have been some federal and state initiatives aimed at protecting publishers from AI, no concrete measures have been implemented, and the current policy silence risks exacerbating the power imbalance between publishers and big tech.

In essence, current regulatory efforts and governance patterns around the world are falling short in supporting journalism and counterbalancing the growing power of platforms. Copyright regimes have consistently prioritized corporate interests and investments over moral rights and individual agency, leaving smaller newsrooms and individual journalists with little negotiating power to fend for themselves.

Compounding the issue, many news organizations have opted for an all-out embrace of AI, hastily introducing AI news anchors and news bots. While these tools may help manage fast-paced news cycles and reduce staff costs, they risk undermining a critical asset: clear, undisputed copyright ownership of news content.

The news industry faces a moment of reckoning, one which will require policymakers to broaden their regulatory approach in a bid to build a healthier information ecosystem. The public, too, must find a way to support quality journalism, or risk seeing it disappear.

Ultimately, the biggest challenge journalism faces may be the erosion of public trust. The stakes of the AI revolution are more than economic. At its core, copyright also encompasses moral rights: It confers authorship, but also accountability. A byline from a trusted journalist with a proven record of quality reporting carries more weight than a bot-generated story.

Editor: Cai Yineng.

(Header image: Visuals from Flying Spark/FY Stock and OlgaZ/VectorStock/VCG, reedited by Sixth Tone)