The digital canvas is alive with a new kind of creation, born not just from human hands but from the intricate dance of algorithms. Generative AI, or AIGC, is rapidly changing how we produce content, and with it, it's posing some pretty fundamental questions for copyright law. At its heart, copyright has always been about protecting human creativity, about recognizing the unique spark of an individual's intellect. But what happens when that spark seems to emanate from a machine?
This is where things get fascinating, and frankly, a little complex. Both China and the United States, as major players in the AI landscape, are grappling with this. While their underlying principles often align – the need to protect intellectual property – their approaches to how they do it show some distinct differences.
The Chinese Approach: Tracing the Human Touch
In China, the cornerstone of copyright protection is "originality." For AIGC, the debate often boils down to the age-old distinction between "idea" and "expression." Some argue that guiding an AI through multiple prompts and refinements is akin to a human artist meticulously selecting and arranging elements, thus constituting a protected expression. Others contend that the prompts are merely ideas, and the AI's output is a random algorithmic product, lacking human authorship.
However, recent court decisions are leaning towards a more nuanced view. Instead of just looking at the final product, courts are increasingly interested in the process. They're asking: did a human have "substantial control" over the final expression? This means looking beyond a simple prompt to examine the detailed refinement, parameter adjustments, and iterative selection that went into the creation. Think of it as moving from a "result-based presumption" to "process-based proof."
We've seen cases where simple, generic prompts lead to outputs that are deemed mere mechanical products of the algorithm, not eligible for copyright. But then there are cases, like the "AI Painting First Case," where detailed prompts, negative prompts, parameter tuning, and multiple rounds of selection were considered evidence of significant human intellectual intervention. The court recognized that this level of human involvement demonstrated "aesthetic choices and personalized expression."
More recently, there's a growing emphasis on the "chain of evidence." It's not enough to just say you guided the AI; you need to show it. This means providing original records of the creation process – prompt versions, comparisons of iterative outputs, editing logs, and timestamped evidence. The "Cat Diamond Pendant Case" is a stark reminder of this, where a copyright claim was rejected because the applicant couldn't provide original creation process records, only post-hoc simulations.
So, in China, the trend is clear: the more granular and demonstrable your intellectual input and control over the AI's output, the stronger your claim to copyright. It's about proving your contribution throughout the entire creative journey.
The American Stance: The "Human Author" Red Line
Across the Pacific, the U.S. copyright system is also focused on originality, but it tends to be more conservative, drawing a firmer line around the concept of a "human author." The core philosophy here is that copyright exists to incentivize human creativity. An AI, being a programmed entity, doesn't need legal or economic incentives to be creative. Therefore, the U.S. Copyright Office generally views AI-generated content as distinct from human authorship.
Their approach often involves a "identify – disclose – strip" logic. If a work contains AI-generated elements, the copyright protection typically extends only to the parts that are demonstrably human-created and identifiable.
When it comes to prompt-based generation, the U.S. Copyright Office often sees it as more akin to "calling a system" rather than a human exercising predictable control over specific expressions. Even highly detailed prompts are sometimes viewed as a form of "gambling" with the AI, where the outcome is determined by the technology itself, not necessarily by a human's unique expressive intent.
The "Théâtre d'Opéra Spatial" case illustrates this. Despite numerous modifications and text prompts, the court didn't find the iterative prompting process to be original expression, highlighting the lack of a unique and deterministic link between human input and AI output.
However, it's not a complete shutout. If you start with human-created material – like a sketch or a draft – and then use AI to refine or calibrate it, the original human expression remains copyrightable. The AI-modified parts, however, are generally not protected. This is often referred to as "rights stripping," where only the "residual" human contribution is recognized.
In practice, this means applicants often need to clearly distinguish between human-created elements and AI-generated ones, and explicitly disclaim copyright over the latter. The "Zarya of the Dawn" case is a good example: while the AI-generated illustrations themselves weren't copyrightable, the human author's original text and the specific arrangement and selection of the illustrations were protected.
Who Owns the Creation?
So, if AIGC can be protected, who is the rightful owner? Under current legal frameworks in both countries, AI itself cannot be a "rights holder." Copyright law typically recognizes natural persons, legal entities, or organizations as authors. AI is seen as a tool, albeit a sophisticated one, not an independent legal entity capable of holding rights or bearing obligations.
In China, if AIGC is deemed copyrightable, the rights usually revert to the human user who provided the intellectual input through prompts and controls, or to the entity that bears legal responsibility for the generation process. Similarly, in the U.S., the focus remains on the human author's original contribution, whether it's the initial creative work or the subsequent arrangement and modification of AI-generated elements.
For businesses looking to leverage AIGC, this means a dual strategy is essential. Documenting the human creative process is key to satisfying the "substantial control" requirements in China. Simultaneously, clearly identifying and delineating human-created elements is crucial for navigating the "human authorship" focus in the U.S. Building a robust system for recording creative input and clearly segmenting human contributions is not just good practice; it's becoming the bedrock of securing intellectual property in this rapidly evolving digital frontier.
