OpenAI once again accused of having used unauthorized content to train ChatGPT

OpenAI can add a new lawsuit to the long list that has occupied its lawyers for years. This time, it is the publisher of Encyclopedia Britannica and its subsidiary Merriam-Webster who are suing the company, accused of shamelessly stealing encyclopedia content without authorization to train its AI models.

Britannica opens a new legal front against OpenAI

According to the lawsuit filed in New York, OpenAI allegedly copied 100,000 articles from Britannica's websites. The publisher also claims that ChatGPT can generate responses almost identical to entries in the encyclopedia and its dictionaries. Internet users no longer need to visit these sites, as the chatbot provides the full answer. Which, for Britannica, is a 'cannibalization' of its web traffic.

Britannica also accuses OpenAI of trademark infringement, particularly when ChatGPT cites the encyclopedia or implies it has authorization to use its content, including in erroneous responses stemming from the AI's 'hallucinations'. OpenAI resorts to the defense used by all major AI players: its models are trained on publicly accessible data, and the use thereof falls under fair use.

Britannica is seeking damages (the amount is not specified) and is asking the court to order OpenAI to cease the unauthorized use of its content. This is not the first lawsuit filed by the publisher against an AI company: since last year, it has also sued Perplexity for similar accusations.

In September, Anthropic proposed a friendly settlement (at $1.5 billion!) to authors to settle a class-action lawsuit and avoid a trial. The arrangement was however rejected by the U.S. court.

👉🏻 Follow tech news in real-time: add 01net to your Google sources, and subscribe to our WhatsApp channel.

Source : Reuters