Newspaper Conglomerate Calls Out OpenAI Training Proposals

A newspaper conglomerate is taking aim at OpenAI’s “absurd” training proposals and calling on the U.S. government to resist “[g]utting generations of copyright protections.”

Technically, said conglomerate refers to two distinct entities, MediaNews and Tribune Publishing. But both those companies are majority-owned by Alden Global Capital, and the appropriate editorial made its way into north of 60 daily newspapers.

Taking a step back for a moment, most are well aware of gen AI giants’ all-encompassing training processes. Leading models have ingested all manner of protected works sans rightsholder permission – with their developers describing the underhanded maneuver as a necessary step towards AI dominance.

Consequently, several of the AI giants are adamant that training on copyrighted materials constitutes fair use. And they’re entreating governments to carve out related training exceptions, which, for obvious reasons, are eliciting criticism in the music world and elsewhere.

Stateside, with the government currently mapping out an “AI Action Plan,” OpenAI last week submitted a 15-page response. In a nutshell, the company not-so-subtly painted a favorable AI regulatory environment as an essential ingredient in the recipe for competitive success against nations including China.

Unsurprisingly, this favorable regulatory environment seemingly refers to implementing relaxed training rules.

“The federal government can both secure Americans’ freedom to learn from AI,” the ChatGPT developer wrote, “and avoid forfeiting our AI lead to the PRC by preserving American AI models’ ability to learn from copyrighted material.”

From there, OpenAI claimed that “rigid copyright rules are repressing innovation and investment” in other markets – and purportedly inhibiting the involved AI operations’ growth in the process.

Enter the newspaper-conglomerate pushback, which roundly attacks OpenAI’s attempt to draw a connection between relaxed copyright rules and national security.

“Gutting generations of copyright protections for the benefit of AI bots would have a chilling effect not just on news organizations but on all creative content creators,” the text reads in part.

On the other hand, “fairly compensating” and obtaining permission from rightsholders “is the right, just and American thing to do,” the document proceeds. “The government should reject these self-serving proposals and protect the work of artists” as well as others yet, per the conglomerate.

Overall, most of the music space appears to agree that training generative AI models without rightsholder permission or compensation is anything but fair. However, it remains to be seen if the courts will feel the same way.

Meanwhile, if OpenAI is right about one thing, it’s that training protections are (and probably always will be) nonexistent in multiple countries around the globe. Especially given AI’s breakneck evolution, it’s unclear how rightsholders can effectively tackle this unprecedented problem in the long term.