The New York Times accused OpenAI of deliberately omitting evidence of artificial intelligence training

Engineers OpenAI Apparently by mistake important evidence that by The New York Times And other publications on AI training data have removed it.

According to The Verge, the legal team of these publications spent more than 150 hours of their time searching OpenAI’s training data. By doing this, they were trying to find data that used New York Times news articles. However, the collected data has been removed in an unspecified manner and it is not even clear what exactly this data consisted of.

OpenAI admitted its mistake and later attempted to recover the data, but the recovered data was “incomplete and unreliable,” according to the filing, which was reported by The New York Times.

While OpenAI’s attorneys call the data erasure a “bug,” The New York Times’ attorneys have noted that there is “no reason to believe” the deletion was intentional.

The legal battle between the New York Times and OpenAI

The battle between The New York Times and OpenAI began in December last year, when the publication claimed that the developer of ChatGPT and its partner, Microsoft, used “millions of articles from this publication” to build their AI tools, and now interacted directly with its content on It is competition.

The New York Times wants OpenAI to accept “billions of dollars in statutory and actual damages” for copying its content.

The publication has reportedly spent more than $1 million so far to fight OpenAI, an expense few publishers can afford. On the other hand, OpenAI has signed deals with major media outlets including Axel Springer and Vox Media (owner of The Verge), suggesting that many publishers would rather partner with it than fight OpenAI.