Elon Musk spent years demanding transparency from everyone else. Twitter files. Government communications. Competitor business models. Now that he's running an AI company, suddenly he wants secrecy. A California judge didn't buy it.

Musk failed to block a California law requiring AI companies to disclose their training data sources. He argued the law would "ruin" xAI by forcing the company to reveal competitive secrets and expose itself to copyright litigation. The judge ruled the public interest in transparency outweighs Musk's business concerns.

The ruling is a major win for transparency advocates - and a headache for every AI company that's been quietly hoping the training data question goes away. It doesn't. Now California-based AI companies will have to document and disclose what copyrighted material they've used to train their models.

Here's the beautiful irony: Musk built his Twitter reputation on demanding transparency. "Sunlight is the best disinfectant," he'd tweet. "People deserve to know the truth." He positioned himself as the free speech absolutist fighting against opaque institutions.

But when California told him to open xAI's books? Suddenly transparency was a threat to innovation. Suddenly disclosure was anticompetitive. Suddenly the public's right to know took a backseat to trade secrets.

The legal arguments in Musk's filing were fascinating in their hypocrisy. He claimed the public "doesn't care" where AI training data comes from - directly contradicting years of public debate about AI companies using copyrighted material without permission. The judge pointed this out, citing extensive media coverage and public interest in exactly this question.

What Musk really fears is what every AI company fears: lawsuits. If xAI has to disclose it trained on copyrighted books, music, art, or code without licensing, creators can sue. And they will. The only reason this hasn't happened at scale is because AI companies have successfully obscured their data sources behind vague terms like