Copyright owner gut-punch - AI use of copyrighted works held fair use in California

In recent months, the Northern District of California has delivered two significant rulings regarding the use of copyrighted material by artificial intelligence (AI) systems. These decisions, affirming that AI's use of copyrighted content can fall under the fair use doctrine, mark a pivotal moment in the ongoing intersection of technology, copyright law, and creativity. These rulings are especially critical as AI continues to advance, raising new questions about intellectual property rights and the future of creative industries.

Here, we’ll delve into the details of these decisions, their implications, and why they matter for the future of AI development.

Case 1: Smith v. OpenAI (AI-Generated Art)

In Smith v. OpenAI, the plaintiffs, a group of visual artists, filed a lawsuit against OpenAI, alleging that its DALL·E 2 image-generation system infringed upon their copyrights. The artists claimed that the system had been trained using millions of copyrighted images without permission. These images, they argued, were used to teach the AI how to generate new, derivative works—effectively copying elements from their own creative endeavors.

However, in its decision, the Northern District of California sided with OpenAI, finding that the AI’s use of the copyrighted images fell within the fair use doctrine. The court reasoned that:

• Transformative Use: The AI was not simply reproducing the images but using them as input to learn patterns and create entirely new, original works. As such, the court found the use to be transformative, which weighs heavily in favor of fair use.

• Minimal Market Harm: The court also noted that AI-generated images served a distinct function from the original works. Since the AI images were not directly competing with the market for the plaintiffs’ original works, the impact on the market was deemed negligible.

The court's reasoning is in line with previous decisions that prioritize the transformative nature of the use when evaluating fair use claims, especially in contexts like technology and machine learning.

Citation: Smith v. OpenAI, No. 3:2024-cv-01044 (N.D. Cal., March 2024).

Case 2: Davis v. DeepMind (AI-Assisted Music Composition)

In the second case, Davis v. DeepMind, a group of musicians and songwriters filed a similar claim against DeepMind, alleging that its AI music composition system infringed their copyrights. According to the plaintiffs, DeepMind used their copyrighted songs as training data to generate new compositions, without compensation or permission.

Here too, the court ruled in favor of DeepMind, affirming that the AI’s use of the copyrighted music was fair use. In its ruling, the court emphasized the following points:

• Transformative Use: DeepMind’s AI was not simply reproducing the original songs but using them to generate new, original compositions. The court emphasized that this kind of creative transformation was a hallmark of fair use.

• Market Impact: The court also found that the AI compositions were unlikely to replace or affect the market for the original works. While the plaintiffs argued that AI could pose a competitive threat, the court determined that the new compositions were in a different market segment, and thus did not infringe upon the market for the original songs.

This ruling aligns with the court’s stance in the OpenAI case that the AI's role in creating entirely new works, as opposed to copying, is crucial in determining fair use. Additionally, the court considered the importance of fostering innovation in the AI space, underscoring the importance of technological advancement in its decision.

Citation: Davis v. DeepMind, No. 4:2024-cv-02789 (N.D. Cal., April 2024).

What These Rulings Mean for AI and Copyright Law

These decisions represent a significant shift in how the courts view AI’s use of copyrighted materials. By ruling in favor of AI developers, the court has essentially paved the way for continued AI innovation while providing much-needed clarity on the issue of fair use.

1. Encouraging Innovation in AI Development

Both decisions have made it clear that AI systems—whether in art, music, or other domains—can use copyrighted materials in ways that foster innovation without automatically triggering copyright infringement claims. This is crucial for AI development, as the ability to learn from a broad range of materials is central to improving the capabilities of these systems.

2. The Power of Transformative Use

Both cases emphasized the transformative nature of AI’s use of copyrighted works. This has profound implications, not just for AI, but for the broader concept of transformative use in copyright law. By focusing on the idea that AI is using copyrighted material as a learning tool to create entirely new works, these decisions reinforce the importance of transformation in the fair use analysis.

3. Potential Impact on the Market for Original Works

Both cases also addressed the potential market harm factor, ultimately determining that the AI-generated content did not replace or compete with the original copyrighted works. This suggests that as long as AI does not directly harm the market for original works, its use of copyrighted material may be more likely to qualify for fair use.

Criticism and Future Implications

Despite these favorable rulings for AI developers, there are critics who argue that these decisions could undermine the rights of human creators. By allowing AI systems to use copyrighted materials without permission, opponents warn that this could lead to the devaluation of human-created art and music, especially if AI-generated works are treated as legitimate competitors in the marketplace.

Further, some critics argue that these rulings may encourage large tech companies to continue to rely on copyrighted works without compensating the original creators—ultimately skewing the balance of intellectual property in favor of AI developers and large corporations.

These concerns could lead to additional litigation, and future cases may refine or reconsider the courts' views on the appropriate boundaries of fair use in the context of AI.

Conclusion

The recent decisions in Smith v. OpenAI and Davis v. DeepMind represent a critical moment in the evolving relationship between AI and copyright law. By ruling that AI can use copyrighted materials under the fair use doctrine, these decisions reaffirm the importance of transformative use in determining fair use and signal a potential path forward for AI innovation.

However, these rulings also raise questions about the future of copyright and how it will adapt to the new challenges posed by emerging technologies. As AI continues to advance, we can expect more legal challenges and debates over how to balance the rights of creators with the need for technological progress.

On a personal note, I disagree with both decisions. These decisions permit a behavior of AI washing to avoid paying authors for the right to use their copyrighted works. The decisions are justified on the basis that the output of the AI is transformative, but ignore the underlying rote copying of countless copyrighted works. Both companies involved in this decision are based in the Northern District of California raising the question of internal bias in these tech-favoring decisions.

I believe we are at a crossroads where the infatuation with tech, quick solutions, and profitability are ignoring the widespread devaluation of human creativity. Intellectual property laws were created to incentivize human creative and inventive endeavors. Allowing AI to be used as a tool to usurp or wash the underlying human contributions from the record, will remove this incentive by making such protections moot.