ChatGPT: Caught in a Copyright Crossfire

Introduction

On December 27, 2023, The New York Times (NYT) launched a legal assault against OpenAI, alleging willful copyright infringement by its generative AI tool, ChatGPT. The accusations center on ChatGPT’s extensive training on NYT articles without authorization and the direct inclusion of NYT content in its outputs. In response, the NYT seeks not only monetary compensation but also the unprecedented remedy of destroying ChatGPT. This drastic request has raised eyebrows and sparked a debate on the legal boundaries of AI development.

The Legal Authority for Destruction

Can a federal court compel the destruction of ChatGPT? The answer, surprisingly, is yes. Copyright law empowers courts to issue destruction orders in certain circumstances. This authority stems from the need to prevent further infringement and eliminate the means by which it occurs. In the case of physical infringing goods like counterfeit records, courts have long ordered their destruction, along with the equipment used in their production.

The Applicability of Destruction to ChatGPT

The NYT argues that ChatGPT fits the criteria for destruction under copyright law. It claims that ChatGPT is an infringing good, similar to pirated vinyl records, as it unlawfully incorporates NYT content. Additionally, the NYT argues that ChatGPT’s training data, which includes NYT articles, constitutes pirating equipment, as it enables the continued infringement.

The Growing Trend of AI-Targeted Legal Actions

While copyright law has not previously been used to destroy AI models, there is a growing trend of legal actions targeting AI systems. The Federal Trade Commission (FTC) has employed algorithmic disgorgement to compel companies to delete unlawfully collected data and AI models trained on such data. This signals a shift toward holding AI developers accountable for copyright violations.

Possible Outcomes of the NYT v. OpenAI Case

Despite the legal authority for destruction and the increasing focus on AI-related copyright issues, it is unlikely that ChatGPT will face the axe. Several more probable outcomes exist:

1. Settlement:

The most straightforward resolution is a settlement between the parties. This would likely result in the dismissal of the lawsuit and avert the destruction of ChatGPT.

2. Fair Use Defense:

OpenAI may assert the copyright doctrine of fair use as a defense. If OpenAI can demonstrate that ChatGPT’s use of NYT content is transformative and does not serve as a substitute for NYT content, it may prevail in its defense.

3. Proving Legitimate Uses or Unnecessary Destruction:

Even if OpenAI loses the case, it may still save ChatGPT by demonstrating that the AI has legitimate, non-infringing uses or that destroying it is not the only way to prevent further copyright violations.

Conclusion

While the NYT’s lawsuit against OpenAI has brought the issue of copyright infringement by AI systems to the forefront, it is highly improbable that ChatGPT will be destroyed. The law provides several avenues for OpenAI to defend itself and preserve ChatGPT. However, this case serves as a wake-up call for AI developers, highlighting the need to address copyright concerns proactively and tread carefully when incorporating copyrighted material into their AI systems.