Generative AI's Use of Artists' Work Sparks Copyright Challenges and Calls for New Compensation Models

Answer first: Generative AI platforms like Stable Diffusion have ingested billions of images and captions from the internet, enabling them to produce new works that mimic artists' styles without legal ownership or compensation. Scholars Kate Crawford and Jason Schultz emphasize that current copyright law, rooted in centuries-old pr…

AI & IP context for IP teams

Generative artificial intelligence systems are reshaping creative production but simultaneously creating a profound copyright crisis for artists. According to a recent article in Issues in Science and Technology, AI platforms such as Stable Diffusion have harvested approximately 5 billion images and text captions, while the Common Pool dataset contains 12.8 billion items. These vast repositories serve as training material for AI models that can then generate new images and texts replicating the artistic styles of original creators.

Kate Crawford, an AI impact researcher at USC Annenberg, and Jason Schultz, a clinical law professor at New York University, explain the implications for artists seeking to earn a living from their work. They note that the billions of outputs produced by generative AI are effectively unowned and can be used by anyone for any purpose. This results in a radical shift in creative production, producing streams of content without any legally recognizable author. As they state, "Whether a ChatGPT novella or a Stable Diffusion artwork, output now exists as unclaimable content in the commercial workings of copyright itself."

Key takeaways for generative AI copyright challenges

• Confirm how the development affects ai & ip ownership, enforcement, licensing, or portfolio records.
• Separate confirmed facts from legal interpretation before advising business teams.
• Map deadlines, affected assets, contracts, and evidence files to the responsible internal owner.
• Use the issue as a prompt for monitoring, filing strategy, dispute preparation, or member education.

Practical analysis

The core problem, Crawford and Schultz argue, is that existing copyright law does not adequately protect individual creators under these new circumstances. Copyright frameworks date back to at least 1710 and were designed for very different technological and cultural contexts. The current legal structure struggles to address AI’s ability to mix, meld, and collate enormous volumes of existing works to produce new content that can stand in for original creation.

Jonathan Bartlett highlighted this issue in December 2023 when the New York Times filed a copyright infringement lawsuit against Microsoft and OpenAI. He observed that while search engines have long used copyrighted content to direct users to original sites—providing at least some benefit to content owners—generative AI and chatbots differ fundamentally. These AI systems use and replace original content without directing users back to the creators, effectively displacing them. OpenAI has acknowledged that its systems violate copyright but relies on legal loopholes to avoid liability.

As litigation against AI companies continues to mount, it becomes clear that generative AI cannot function without access to these billions of images and texts. This reality presents a stark choice: either forego such AI technologies or develop mechanisms to compensate the original content producers.

Robert J. Marks, writing for Newsmax in January 2024, points to existing models for compensating creators. He cites Spotify’s system, which automatically tracks song plays and distributes royalties accordingly. Marks suggests similar methods could be adapted for generative AI content, but this would require transparency and record-keeping from AI system operators, who currently maintain secrecy around their data and processes.

Marks also notes that end users will ultimately bear the cost of using AI services. The reason chatbots are often free while traditional media like Disney movies require payment is partly because AI vendors have not compensated the original creators whose work underpins the AI outputs. Without fair payment, creators may cease producing new content, leading to what is known as the "jackrabbit problem" or model collapse—where AI output deteriorates due to repetitive regurgitation of the same material.

The sustainability of AI-generated content depends on continuous input from human creators. If artists and writers are not remunerated, the quality and diversity of AI outputs will decline, undermining the technology’s long-term viability.

In sum, the challenges posed by generative AI to copyright law are unlikely to be resolved by applying existing legal frameworks. Instead, a new system is needed—one that recognizes the unique nature of AI technologies and establishes fair compensation schemes for the producers of the material AI uses.

Additional commentary on this issue includes concerns about "cyber plagiarism," where AI systems appropriate copyrighted images without authorization, potentially exposing AI companies to legal jeopardy. Furthermore, the fundamental operational problem of model collapse underscores the necessity of ongoing human creative input, which cannot be replaced solely by algorithms.

As the legal landscape evolves, stakeholders must balance innovation with the rights and livelihoods of original creators to ensure a fair and sustainable future for creative industries in the age of AI.

Related IIPLA reading

• Latest IP news
• IP resources library
• IP guide
• IIPLA events and webinars