Claude users are canceling their subscriptions, citing Anthropic’s new AI watermark

Anthropic's move to watermark Claude-generated text is driving immoderate users distant from its AI assistant.

The AI laboratory said this week that immoderate versions of Claude will embed an "imperceptible watermark" straight into their matter output. It will recreation pinch the contented erstwhile it is copied and pasted and whitethorn past immoderate editing, Anthropic said, helping to separate AI-generated worldly from quality writing.

Anthropic said watermarking does not alteration the meaning, quality, aliases readability of Claude's response. Despite that, dozens of group person posted connected X since Monday that they had canceled their Claude subscriptions, citing the watermark.

Business Insider said pinch 4 users who said they had canceled their Claude subscriptions. Some users, who shared confirmation of their cancellations pinch us, said they disliked the imaginable of a persistent marker successful their activity and were worried astir their AI usage showing up successful customer work, specified arsenic coding.

While it's excessively early to opportunity what the effect could beryllium for Anthropic — the institution had 300,000 business customers arsenic of September 2025, a number wide expected to person grown importantly since past — backlash to the watermark shows nary motion of slowing.

Anthropic told Business Insider that it hasn't seen a inclination of an uptick successful cancellations since it announced the watermark.

Why users opportunity they're canceling

Arturo Villarroya, who useful successful nationalist policy, told Business Insider he canceled his Claude Max subscription — a work that starts astatine $100 a period — connected Wednesday aft becoming "uncomfortable" pinch Anthropic's caller watermarking plan. It has been applied globally to text from supported Claude models since August 2.

"I do usage it for everything, mostly coding apps and tools," he said of Claude, adding that he besides utilized it arsenic a support instrumentality for sourcing, research, and corrections astatine work.

Villarroya said he's worried a watermark could stay moreover erstwhile the underlying penning was original and Claude was utilized only to translate, cheque spelling, hole world citations, aliases lightly summarize a passage.

He said its beingness could create problems successful world and master settings pinch strict rules connected AI use. He added that he is switching to Cursor and Grok 4.6 for coding, and Grok Bot for activity and individual organization.

Anthropic said successful its blog that a detected watermark indicates Claude processed text, not that it was needfully the original author. The institution said marks tin stay aft Claude proofreads, translates, summarizes, aliases different edits content.

That favoritism did not easiness concerns for some. Vladislav Rajtmajer, a freelance developer successful the Czech Republic, told Business Insider he canceled his Claude Max subscription connected Tuesday, citing the watermark arsenic the main reason.

He said he uses AI for codification reviews and translations, and worried an AI explanation connected codification shipped to clients could raise questions astir authorship aliases trigger statement penalties.

He added that the watermark was a "signal to look astatine Chinese AI providers and study to activity pinch different models."

'They adhd the watermark anyway'

Richard Echols, an AI advisor successful Georgia, besides said he canceled his Claude Max subscription connected Wednesday and that the watermark was a awesome facet successful his decision.

He said he uses Claude to create archiving for businesses and worried the people could look erstwhile he had written the underlying worldly himself and utilized the chatbot only for edits.

"Even if you edit your ain activity that you wrote, they adhd the watermark anyway," Echols said.

Anthropic said successful its blog that a watermark shows that Claude processed text, not needfully that it wrote it, and that the people tin stay aft Claude has proofread, translated, aliases summarized content.

Stevan Bogosavljević, a elder package technologist successful Montenegro, canceled his Claude Max subscription successful July, citing what he views arsenic declining work quality, constricted entree to Anthropic's apical models, and stronger alternatives. He said the watermark further validated his determination to cancel.

Having codification attributed to Claude "definitely would beryllium not desirable," he said, arsenic he would not want professionally shipped codification to transportation an invisible marker that could raise questions astir authorship, compliance, aliases customer policies.

The lawsuit for AI watermarks

Not everyone is against the usage of AI watermarks, pinch immoderate saying they amended transparency arsenic much online contented is generated by artificial intelligence. Others person said they could thief forestall AI systems from being trained connected a increasing measurement of AI-generated material, which tin origin systems to degrade.

Aadit Sheth, cofounder of executive communications patient The Narrative Company, said connected X connected Tuesday that audiences should beryllium capable to show whether the words they publication bespeak a person's ain reasoning and that watermarks are 1 measurement to guarantee that.

Anthropic isn't the only AI laboratory to usage watermarks: Google uses its SynthID exertion to watermark AI-generated content, and OpenAI uses SynthID for supported images and audio.

Elon Musk's societal media level X besides adds a "Made pinch AI" tag connected contented it determines to beryllium AI-generated aliases manipulated. The tag appeared connected the resignation post of outgoing White House property caput Karoline Leavitt connected Wednesday, but later disappeared.

Anthropic said that it added the watermark successful bid to comply pinch the European Union's AI Act, and that it plans to rotation retired a free exertion programming interface, aliases API, that will let users and 3rd parties to cheque matter for Claude's watermark themselves.

Santiago Alessandro Rivera Martínez, a package technologist and laminitis of the integer agency Directa, said Anthropic's watermark reinforced his concerns of becoming excessively limited connected Claude. He said that had led him to cancel his subscription this week.

"When your workflow is built wholly astir proprietary software, the supplier tin alteration the rules, features, restrictions, aliases conditions astir that workflow later," he said.

"I would alternatively person the AI furniture of my activity beryllium portable and provider-agnostic," he added.