Gruber calls Anthropic's plan to make Claude's word choices traceable "a perversion of writing." The company argues watermarking does not degrade output quality.
Alpha Score of 57 reflects moderate overall profile with strong momentum, poor value, strong quality. Based on 3 of 4 signals – score is capped at 90 until remaining data ingests.
Tech blogger John Gruber is the latest critic to take aim at Anthropic's plan to watermark text generated by its Claude AI model.
On Sunday, Gruber, who writes the popular blog Daring Fireball, called the watermarking approach a "perversion of writing." His central argument: Claude should pick words because they serve the user, not because they make the output traceable.
Anthropic has described the feature as an "imperceptible watermark" embedded in Claude's text. The plan emerged from the European Union's AI Act, which requires machine-readable marking on all AI-generated content. Anthropic wrote that other major models will eventually need similar systems.
In a Friday post headlined "How Claude's text watermark works," the company said the method "does not impact the quality of Claude's output" and does not rely on hidden characters. Instead, Claude's word-selection process gets adjusted in subtle ways to create a statistical pattern. Anthropic argues those adjustments apply only to words that are virtually interchangeable in context.
The company gave an example: if Claude writes "The weather today was cold and…" the model is likely to pick "gray" or "overcast" next. The sentence's meaning does not change either way, Anthropic claims.
Gruber is not persuaded. In his view, once watermarking alters word choice at all, something other than producing the best possible prose is steering the result.
"The exact words we choose when writing matter," he wrote. "I want any LLM I use to choose the very best, most precise words at every single decision point."
The disagreement highlights the confusion around Anthropic's plans. Some Claude users have cancelled their subscriptions over the feature, though the company told Business Insider it has not seen a spike in cancellations. Anthropic did not immediately respond to a request for comment.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.