Authors to Review OpenAI Training Data in Secure Room Amid Copyright Lawsuit

Authors Suing OpenAI Will Review Its Training Data 

But Only in a Highly Guarded Room

Authors suing OpenAI for copyright infringement will review its confidential training data under strict rules. Learn what this means for AI and copyright law.

In what sounds more like a spy movie than a copyright lawsuit, writers suing OpenAI for copyright infringement will soon gain access to the company’s confidential training data, but under extremely strict conditions.

Writers Allowed Access, With Heavy Restrictions

Attorneys representing authors Sarah Silverman, Ta-Nehisi Coates, and Paul Tremblay confirmed in a court filing this week that they reached an agreement with OpenAI. According to The Hollywood Reporter, this deal grants their legal team access to OpenAI’s training dataset for review.

However, OpenAI isn’t making it easy.

The plaintiffs’ agents can only examine the data inside a secure room at OpenAI’s San Francisco headquarters. The room will house a single locked-down computer that’s disconnected from the internet and all shared networks.

No Copies, No Phones, Just Notes

Visitors can take handwritten notes while viewing the data, but OpenAI prohibits any digital copying or the use of external devices. No phones, cameras, or laptops are allowed. Even though the AI was trained on public content, OpenAI is treating this like state-level secrecy.

Before entering the room, every person must sign a non-disclosure agreement (NDA), present official identification, and log their names in a visitor register.

A Legal Battle With Big Implications

This legal clash could shape how AI companies use copyrighted content going forward. Coates, Silverman, and Tremblay are represented by the Joseph Saveri Law Firm, which also leads a similar lawsuit against Meta.

The writers argue that OpenAI used their copyrighted works to train ChatGPT without consent or compensation. As a result, the chatbot now generates responses that allegedly plagiarise their intellectual property.

While some claims in the lawsuit were thrown out earlier this year, including those involving negligence and unjust enrichment,  U.S. District Judge Araceli Martinez-Olguin allowed the core copyright infringement claims to proceed.

When Will the Review Happen?

So far, the timeline remains unclear. We don’t yet know when the authors’ legal representatives will inspect the data, how long they’ll have access, or how many people will be allowed in the room at one time.

One thing is certain: the outcome of this lawsuit could set a critical precedent for the future of AI development. Whether or not the plaintiffs find a “smoking gun” inside OpenAI’s dataset, this case will likely influence how tech companies handle copyright in the age of artificial intelligence.