Throughout the two-year legal battle, OpenAI maintained that scanning its massive training corpus or ChatGPT conversation logs was technically impossible and posed severe privacy risks. However, a court-ordered deposition of data privacy engineer Vinnie Monaco suggests otherwise. Monaco revealed that the company had already conducted internal evaluations to track the use of copyrighted journalism within its models.
In section Startups & Technology
New York Times Accuses OpenAI of Concealing Copyright Evidence
The New York Times and The Daily News are petitioning a judge to sanction OpenAI, alleging the firm systematically withheld evidence and manipulated discovery in their ongoing copyright lawsuit. Plaintiffs claim OpenAI falsely asserted it lacked the technical capacity to audit its training data while secretly maintaining internal logs of copyrighted content.
Evidence suggests that prior to the lawsuit, OpenAI compiled a database of 78 million de-identified conversations to monitor potential infringement. Following the legal filing, the company reportedly implemented "Project Giraffe," utilizing a "Bloom" filter to detect and record instances where the AI regurgitated copyrighted material. Plaintiffs argue that OpenAI intentionally provided an "unusable," heavily redacted sample of 20 million chat logs while allegedly deleting billions of records in violation of a preservation order. Lead counsel Ian B. Crosby stated that if the company believed its practices were legal, it would not have obscured the truth regarding these internal audits. OpenAI spokesperson Drew Pusateri dismissed the allegations as "blatantly false," accusing the media outlets of attempting to invade user privacy as their underlying legal strategy falters.
Comments (0)
No comments yet. Be the first!