The New York Times and The Daily News have accused OpenAI of hiding evidence in their ongoing copyright lawsuit, revealing that OpenAI had tools to search for copyrighted journalism in ChatGPT outputs. The plaintiffs are now seeking sanctions against OpenAI for allegedly withholding crucial data.
TL;DR
- OpenAI is accused of hiding tools and datasets that could identify copyrighted journalism in ChatGPT outputs.
- The plaintiffs, including The New York Times and The Daily News, are seeking sanctions against OpenAI for allegedly withholding evidence.
- OpenAI denies the allegations, stating that the Times is attempting to access private user conversations.
What happened
The New York Times and The Daily News claim that OpenAI has been lying about its ability to search customer chat log data and training datasets for their copyrighted works. This is part of a two-year lawsuit against OpenAI for allegedly violating copyright law by training its generative AI models on the Times' content and reproducing that journalism in user outputs.
During a court-ordered deposition in April, OpenAI data privacy engineer Vinnie Monaco allegedly revealed that OpenAI had already conducted internal searches and evaluations of its training corpus to search for copyrighted journalism works. Monaco's deposition also allegedly revealed that OpenAI had amassed a database of about 78 million de-identified ChatGPT conversations and implemented a 'Bloom' filter as part of a set of tools called 'Project Giraffe' to detect and record regurgitation in outputs.
The plaintiffs originally asked OpenAI to provide a sample of 120 million chat logs, but OpenAI negotiated to bring the sample down to just 20 million. However, the plaintiffs claim that the sample submitted by OpenAI was heavily redacted and unusable. They also allege that OpenAI deleted billions of ChatGPT outputs after they filed suit and substituted millions of logs in the requested sample.
Why it matters
This escalation in the lawsuit could have significant implications for OpenAI and the broader AI industry. If the court finds that OpenAI withheld evidence, it could face sanctions and be prevented from using the 20 million chat log sample as evidence. This could weaken OpenAI's defense and potentially lead to a ruling in favor of the plaintiffs.
For developers and startups, this case highlights the importance of transparency and compliance with copyright laws when training AI models. It also underscores the potential legal risks associated with using copyrighted material in AI training datasets.
For investors, this legal battle could impact OpenAI's valuation and reputation. If OpenAI is found to have violated copyright laws, it could face significant financial penalties and damage to its brand, which could deter potential investors.
Key facts
- OpenAI is accused of hiding tools and datasets that could identify copyrighted journalism in ChatGPT outputs.
- The New York Times and The Daily News are seeking sanctions against OpenAI for allegedly withholding evidence.
- OpenAI data privacy engineer Vinnie Monaco allegedly revealed that OpenAI had conducted internal searches and evaluations of its training corpus.
- OpenAI had amassed a database of about 78 million de-identified ChatGPT conversations and implemented a 'Bloom' filter as part of a set of tools called 'Project Giraffe'.
- The plaintiffs originally asked for a sample of 120 million chat logs, but OpenAI negotiated to bring it down to 20 million.
- The plaintiffs claim that the sample submitted by OpenAI was heavily redacted and unusable.
- OpenAI denies the allegations, stating that the Times is attempting to access private user conversations.
Context
This lawsuit is part of a broader trend of legal challenges facing AI companies over the use of copyrighted material in training datasets. As AI models become more advanced, they require vast amounts of data to train, often leading to conflicts with copyright holders.
The outcome of this case could set a precedent for future legal battles involving AI and copyright law. It could also influence how AI companies approach data collection and training in the future, potentially leading to more transparent and compliant practices.
For the AI industry, this case highlights the need for clear guidelines and regulations regarding the use of copyrighted material in AI training. It also underscores the importance of balancing innovation with respect for intellectual property rights.