Μake us preferred on Google When American investor Stanley Druckenmiller’s essay landed in Paul Gigot’s inbox this week, the WSJ’s opinion editor was pleased. Here was a storied investor, someone who had once worked alongside Treasury Secretary Scott Bessent and Fed governor Kevin Warsh, arguing that the Treasury shouldn’t be trying to control bond prices. It was exactly the kind of piece the opinion desk hopes will make waves.
And it did. Financial outlets around the globe picked it up, Wall Street took notice, and it’s a safe bet Bessent himself read every word.
But attention quickly shifted away from Druckenmiller’s argument and toward the media’s favorite topic: itself. Rumors spread online that AI had written the piece, fueled by a growing crop of tools that claim to spot machine-generated text, tools whose reliability is itself questionable. When a reporter asked Druckenmiller directly, he admitted he’d used a large language model, comparing it to using a calculator, and noting that he isn’t a professional writer.
The reaction was swift outrage. Commentators recoiled at the idea that a wealthy public figure hadn’t personally written every sentence himself. Journalists then pressed Gigot for comment, hoping he would either distance the paper from Druckenmiller or concede that publishing the piece had been a mistake. He did neither, which only intensified the criticism. Media critics called the decision reckless, warning it opened the door to a flood of low-quality AI-generated content across the industry.
Gigot argues this reaction misunderstands both how AI is actually used today and what an opinion section is for. He notes that using AI for research, grammar checks, or editing help has become commonplace, and that treating this as a simple yes-or-no question of “using AI” oversimplifies the matter. He points to Washington Post columnist Megan McArdle, who has said publicly that she uses AI as a sounding board, similar to talking an idea through with a colleague, and suspects many other writers quietly do the same.
Still, Gigot draws a firm line for his own staff: their columns and editorials must be entirely original work, with no AI drafting permitted, so readers can trust that both the ideas and the writing are genuinely theirs. He adds that part of what makes reading a columnist like Holman Jenkins worthwhile is following his distinctive personal style.
Outside contributors, he says, are a different matter. People like Druckenmiller aren’t professional writers, but their expertise and public standing give their opinions real value. Gigot notes that many prominent contributors, politicians and executives especially, already rely on speechwriters or professional writing help, and he doesn’t see using AI to polish an essay as fundamentally different.
What ultimately matters to Gigot is whether the argument belongs to the person putting their name on it. Druckenmiller’s views on the bond market and federal deficits, he says, are worth hearing regardless of how the prose was assembled, given they’re backed by decades of market experience. Druckenmiller has written for the Journal before, likely with help from staff, and this time turned to AI instead. In both cases, Gigot maintains, the underlying ideas were his own.
As for calls to disclose AI use in submissions, Gigot is skeptical of what that would even mean in practice, questioning whether outlets would need to interrogate every contributor about exactly how they used a chatbot. He acknowledges this episode may tempt some writers to submit low-effort, AI-generated work, and says his team will keep trying to catch and reject it as they always have. But he won’t pretend that contributors never use AI tools. Druckenmiller’s real offense, in Gigot’s view, was simply admitting it, and he says the investor’s contribution to the debate remains valuable and welcome in the paper’s pages going forward.
Information is based on WSJ and The Atlantic
