Scoop: Nielsen's Gracenote sues OpenAI for copyright infringement
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Natalie Peeples/Axios
Gracenote, a metadata and identification services company owned by media measurement giant Nielsen, has sued OpenAI for copyright infringement, Axios has learned.
Why it matters: Gracenote is suing OpenAI not just for using its metadata without authorization or compensation, but also for copying the relational framework it uses to connect its metadata, which is in part what makes the data valuable to its enterprise clients and useful for consumers.
- To date, there hasn't been a major media copyright lawsuit that focuses on the theft of a proprietary sequence or structure behind a dataset.
- This lawsuit could set a new precedent for how data providers, in the media industry and outside of it, protect their intellectual property.
Zoom in: The complaint, filed by Susman Godfrey lawyers on behalf of Nielsen's Gracenote in the Southern District of New York, alleges OpenAI copied and used Gracenote's data for its large language models that support lucrative products like ChatGPT.
- The plaintiffs argue outputs from OpenAI's products contain and generate exact copies of Gracenote's data and relational framework, which includes human-curated and analyzed sets of entertainment metadata across music, video and sports.
- The complaint alleges the unauthorized use of that data and the relational framework that connects it — both of which are copyright protected — threaten Gracenote's core business with media content distributors, such as smart TV providers.
- Those providers could use the data and framework scraped by OpenAI to build their own "substitutive, competing media metadata products and platforms, all without permission or compensation to Gracenote," it reads.
How it works: Gracenote employs hundreds of editors who use human insight and judgment to create millions of narrative descriptions, original video descriptors, unique identifiers and other program identifiers that TV providers and other clients can use to help customers discover content.
- For example, Gracenote editors described HBO's "Game of Thrones" as "the depiction of two power families — kings and queens, knights and renegades, liars and honest men — playing a deadly game of control of the Seven Kingdoms of Westeros, and to sit atop the Iron Throne."
- In the lawsuit, Gracenote alleges OpenAI scraped and used a near-exact copy of that descriptor when prompted by a ChatGPT user to describe "Game of Thrones."
- It provides several other examples where, with minimal prompting, OpenAI's various ChatGPT models recite large portions of Gracenote's program descriptions verbatim.
Between the lines: Gracenote's entire Programs Database, which includes its metadata and the proprietary relational map its editors use to connect that data, is registered with the U.S. Copyright Office.
- Because of those protections, the company is suing OpenAI for statutory damages, in addition to actual damages.
- Statutory damages are preset amounts for specific legal violations, in this case copyright infringement. Actual damages are awarded to cover losses of a legal violation.
The other side: A spokesperson for OpenAI said in a statement to Axios, "Our models empower innovation, and are trained on publicly available data and grounded in fair use."
Zoom out: Gracenote's complaint is notable given the fact that the company has expressed willingness to work with AI companies, and has struck AI licensing deals with firms like Samsung and Google.
- Gracenote says in its lawsuit that it reached out to discuss licensing its data to OpenAI "many times over an extended time period" but said the AI company "rebuffed or ignored every single attempt to do so."
- "Being pro-AI and anti-theft aren't contradictory; they are the only sustainable path forward. We've filed suit to protect that future," Gracenote CEO Jared Grusd said in a statement.
The big picture: The are many pending cases between media and information companies, and AI firms that could impact how a judge evaluates Gracenote's lawsuit and others.
- AI companies that scraped online materials to train their models tend to argue that public data is "fair use" under existing copyright law, in part because that data can be used to create new and helpful services or information for consumers.
- The Gracenote lawsuit argues OpenAI's outputs are near-exact copies of the metadata it licenses to clients, which means that the information OpenAI is providing to consumers it isn't necessarily new.
Disclosure: Axios and OpenAI have a licensing and technology agreement. Axios has editorial independence.
Editor's note: This story was updated to include comment from OpenAI and to include the above disclosure.
