On June 25, 2025, the United States District Court for the Southern District of New York was seized of a new copyright infringement lawsuit filed by Jeff Bird, a professional photographer and author based in New York City, against Microsoft Corporation (Case No. 1:25-cv-05282).
This action adds to the growing wave of litigation in the United States surrounding generative artificial intelligence (AI) and the unauthorized use of copyrighted works to train large language and multimodal models. Jeff Bird is the registered copyright owner of multiple photographic works, including his acclaimed collection “Bird’s Eye” (U.S. Copyright Reg. No. VA 2-183-382), and alleges that his photographs were used without consent to train OpenAI’s GPT models, which are embedded in Microsoft’s AI offerings such as Copilot, Bing Chat, and Azure OpenAI Services.
The complaint is founded on several core provisions of Title 17 of the United States Code, including:
Bird seeks to hold Microsoft directly and secondarily liable, arguing that its collaboration with OpenAI constitutes “a joint enterprise” in which Microsoft “knowingly provided the infrastructure, funding, and commercial integration” of models that were “trained on massive datasets containing copyrighted works, including Plaintiff’s.”
The photographs at issue consist primarily of high-resolution, original images of natural landscapes, wildlife, and aerial views, which are described in the complaint as:
“carefully curated photographic compositions known for their technical sophistication, aesthetic detail, and artistic value.”
The works were initially published in fine art books, exhibition catalogs, and licensed for editorial and commercial use. Bird alleges that his images were scraped from online platforms and image repositories and incorporated into training datasets without authorization, attribution, or compensation.
The allegedly infringing uses include:
According to the complaint:
“Defendants’ AI products were trained on datasets containing Plaintiff’s copyrighted works, enabling them to generate outputs that compete with, and in some cases closely replicate, the original photographs.”
“Microsoft’s integration of these models into its commercial ecosystem knowingly leveraged this unauthorized use for profit.”
The plaintiff argues that Microsoft cannot avail itself of the fair use defense, stressing that the use of copyrighted works to train large-scale AI systems for commercial purposes is neither transformative nor justified:
“Defendants cannot rely on the fair use doctrine to justify the massive ingestion and use of copyrighted works for commercial AI training purposes, especially when the resulting outputs are themselves capable of substituting for the original works.”
The complaint also insists that:
“Microsoft’s use of Plaintiff’s works is not transformative in any legally relevant sense—it merely exploits the works to build a tool that can mimic or replace them.”
Bird further contends that Microsoft acted with knowledge or reckless disregard for the infringement:
“Microsoft knew or willfully ignored that the AI models it was funding and deploying were trained on copyrighted works without permission.”
“Microsoft was not a passive downstream user—it played an active role in the design, development, funding, and commercial deployment of the infringing models.”
Bird is seeking:
The complaint also reserves the right to amend claims as further evidence emerges regarding Microsoft’s role in the training pipeline.
This case is part of a broader legal offensive in the United States against the opaque practices of AI model training, which has seen a surge of lawsuits by authors, illustrators, journalists, and publishers since 2023.
It underscores the stark contrast between the rapid legal mobilization in the U.S. and the near absence of equivalent litigation in France and continental Europe, where no major decision on AI training and copyright has yet been issued.
The Bird v. Microsoft complaint illustrates the legal pressure mounting on major tech firms and raises urgent questions for rightsholders worldwide. It calls into question whether existing copyright frameworks are adequately equipped to confront the scale and complexity of AI development—and whether European authors will be able to assert their rights with equal force.
This article was prepared by a lawyer admitted to the Paris Bar and is provided for general informational purposes only. For any legal advice concerning U.S. copyright law or proceedings before a U.S. federal court, readers are advised to consult a qualified attorney licensed to practice in the United States.
