OpenAI’s copyright case gets federal backup

The US Government just chose a side in the copyright fight.

In an industry-shaking move, the Justice Department argued that OpenAI can legally train its AI on The New York Times’ reporting. Translation: AI developers may be free to scale their models on publishers’ work. No permission or paycheck required.

But the DOJ didn’t stop there. It also claimed that using The Times’ articles to teach AI systems was vital to national security. A classic case of the end justifying the means. 

All quips aside, there’s a major dilemma here around copyright and transformation. This was seen in Andy Warhol v. Goldsmith, which proved a work can look new and still fail the fair-use test if it serves the same commercial purpose as the original.

Now that same question hangs over generative AI. Is AI training on publishers' work then turning around and undercutting its competition?

This is (in part) what The Times is rebutting. “The administration’s proposal to let companies take that content without permission or compensation would undermine the sustainability of the human-created content that a healthy society depends on, and which A.I. needs to function”

The copyright conflict

Word from the DOJ holds weight, but it’s not the final say. Judge Sidney H Stein will have to consider the whole messy equation: copyright, competition, compensation. And how much collateral damage Washington is willing to accept in the global AI race. 

Even still, the battle ain’t over until final the gavel drops. This case could end with an independent ruling, or begin crawling its way up the ladder toward the Supreme court. 

Whether the final byline goes to a Times reporter or an AI agent remains TBD.

AI wants to train its cake and eat it too. - TL