Anthropic’s move to watermark Claude-generated text is turning some users away from its AI assistant. The AI Lab said this week that some versions of Claude will incorporate an “imperceptible watermark” directly into its text output. It will travel with the content when copied and pasted and can survive a few edits, Anthropic said, helping distinguish AI-generated material from human writing. Anthropic said the watermark does not change the meaning, quality or readability of Claude’s answer. Despite that, dozens of people have posted on X since Monday that they had canceled their Claude subscriptions, citing the watermark. Business Insider spoke to four users who said they had canceled their Claude subscriptions. Some users, who shared confirmation of their cancellations with us, said they didn’t like the prospect of a persistent marker in their work and were worried that their AI use would show up in client work, such as coding. While it’s too early to say what the impact could be for Anthropic – the company had 300,000 business customers in September 2025, a number that is expected to have grown significantly since then – the reaction to the watermark shows no signs of slowing down. Business Insider has not seen an increasing trend in cancellations since announcing the watermark. Why Users Say They’re Canceling Arturo Villarroya, who works in public policy, told Business Insider that he canceled his subscription to Claude Max, a service that starts at $100 a month, on Wednesday after feeling “uncomfortable” with Anthropic’s new watermarking plan. It has been applied globally to the text of supported Claude models since August 2. “I use it for everything, mainly applications and coding tools,” he said of Claude, adding that he also used it as a support tool for sourcing, research and corrections at work. Villarroya said he is concerned that a watermark could remain even when the underlying writing was original and Claude was used only to translate, check spelling, correct academic citations or slightly summarize a passage. He said its presence could create problems in academic and professional settings with strict rules on the use of AI. He added that he is switching to Cursor and Grok 4.6 for coding, and Grok Bot for work and personal organization. Anthropic said on its blog that a detected watermark indicates text processed by Claude, not necessarily the original author. The company said brands can remain after Claude proofreads, translates, summarizes or edits the content. That distinction did not alleviate the concerns of some. Vladislav Rajtmajer, an independent developer in the Czech Republic, told Business Insider that he canceled his subscription to Claude Max on Tuesday, citing the watermark as the main reason. It said it uses AI for code reviews and translations, and was concerned that an AI label on code sent to customers could raise questions about authorship or trigger contractual penalties. He added that the watermark was a “signal to look at Chinese AI vendors and learn to work with other models.” “They add the watermark anyway,” Richard Echols, an AI consultant. in Georgia, also said he canceled his Claude Max subscription on Wednesday and that the watermark was a major factor in his decision. He said he uses Claude to create documentation for companies and is concerned that branding could appear when he wrote the underlying material himself and used the chatbot only for editing. “Even if you edit your own work that you wrote, they add the watermark anyway,” Echols said. Anthropic said on its blog that a watermark shows that Claude processed the text, not necessarily that he wrote it, and that the mark may remain after Claude has revised, translated or summarized the content. Stevan Bogosavljević, a senior software engineer in Montenegro, canceled his subscription to Claude Max in July, citing what he sees as declining service quality, limited access to Anthropic’s best models, and more robust alternatives. He said the watermark further validated his decision to cancel. Having code attributed to Claude “would definitely not be desirable,” he said, as he wouldn’t want professionally submitted code to carry an invisible marker that could raise questions about authorship, compliance or client policies. The Case for AI Watermarks Not everyone is against the use of AI watermarks, and some say they improve transparency as artificial intelligence generates more content online. Others have said they could help prevent AI systems from being trained on an increasing volume of AI-generated material, which can cause systems to degrade. Aadit Sheth, co-founder of executive communications firm The Narrative Company, said on X on Tuesday that audiences should be able to tell whether the words they read reflect a person’s own thinking and that watermarks are a way to ensure that. Anthropic isn’t the only AI lab using watermarks: Google uses its SynthID technology to mark AI-generated content, and OpenAI uses SynthID for compatible images and audio. Elon Musk’s X social media platform also adds a “Made with AI” label on content that it determines is generated or manipulated by AI. The label appeared in outgoing White House press secretary Karoline Leavitt’s resignation message on Wednesday, but then disappeared. Anthropic said it added the watermark to comply with the European Union’s AI Law and that it plans to implement a free application programming interface, or API, that will allow users and third parties to check the text themselves for Claude’s watermark. Santiago Alessandro Rivera Martínez, software engineer and founder of digital agency Directa, said Anthropic’s The Watermark reinforced his concern about becoming too dependent on Claude. He said that had led him to cancel his subscription this week. “When your workflow is based entirely on proprietary software, the vendor can change the rules, features, restrictions or conditions around that workflow later,” he said. “I would prefer the AI layer of my work to be portable and vendor-agnostic,” he added.
