GoDaddy's CEO Trashes AI-Written Documents: 'I Don't Read It'

GoDaddy CEO Aman Bhutani refuses to read any document he suspects was written by AI. And he makes the call within seconds.

"If somebody sends me a document and I can tell pretty quickly it was written by AI, I don't read it," Bhutani told Business Insider in a new interview. "Because if the person couldn't take the time to write it, then I definitely shouldn't take the time to read it."

The stance puts the head of a $15 billion company — the world's largest domain name provider — directly against the most popular workplace use of the technology. In Gallup's May 2026 survey, 51% of workplace AI users said they used the technology for writing and editing, making those tasks the most common AI applications among U.S. workers.

Bhutani isn't banning AI entirely. He draws a line between using the technology as a clarifier and letting it take over the writing. An employee can turn to AI to make sense of an unfamiliar word or phrase, he said, but they must still convert their own ideas into a clear narrative.

The takeaway: at GoDaddy, AI is a dictionary, not a ghostwriter.

What is Bhutani actually looking for?

Original thought. Bhutani said writing documents yourself builds what he calls "the art of iteration" — refining your written work and improving over time. Skip that practice, and the skill atrophies.

"What I'm really looking for is: What was the original observation? What was the original hypothesis, and what got you interested in this idea? Why do you think it matters? Right?" he said. "Because AI only knows what the internet already knows. It doesn't know what we know deep inside us, and there are ideas coming from us all the time."

Turning ideas into compelling narratives is a critical skill, Bhutani added. That's the standard his documents get measured against.

In plain words: the CEO grades the thinking, not the polish.

How does he spot AI-written text?

Bhutani relies on a few tells, and one of them is personal to his workforce. Many of his colleagues speak English as a second language, so unusually polished prose from them raises his suspicion immediately.

Length works both ways. Some AI-generated pieces run too long. Others come back short but carry a recognizable "punchy and clear" tone that marks AI.

Research backs him up. Researchers at Florida State University have found that common words such as "delve," "intricate," "realm" and "underscore" can signal AI authorship, especially when those keywords appear repeatedly.

The stylistic fingerprints go beyond vocabulary:

• Repeated contrast formulas, such as "It's not just X — it's Y."
• Flowery metaphors, like describing a subject as an "intricate tapestry."
• Perfect grammar and a plethora of em dashes.
• Overly uniform structure: similarly sized sections, repeated lists, and conclusions that restate the introduction.
• Inflated language, with grand claims and elaborate wording.

Each clue is a signal, not proof. Detection tools can misfire too.

Can AI detection be trusted?

Not by itself. AI detection tools like Turnitin can misidentify human writing as machine-generated. Turnitin itself advises against using its results alone to take action against a writer.

That caution matters for any manager tempted to copy Bhutani's approach. Even a CEO who says he can tell "pretty quickly" is working from pattern recognition, not verified fact — and the tools built to confirm his hunches carry documented error rates.

The bottom line: as AI writing spreads, the scarce workplace skill shifts from clean prose to demonstrable original thinking.