Users abandon Anthropic Claude over invisible watermarks in generated text

Anthropic announced this week that it will start adding 'invisible watermarks' to text generated by some versions of its Claude AI assistant. The watermarks persist through copy-paste and sometimes even editing, but the company says they won't affect the meaning, quality, or readability of responses. Despite this, dozens of users have taken to social media to report cancelling their paid Claude subscriptions, put off by the prospect of permanent watermarks in their work, fearing it would become obvious that they used AI when dealing with clients or writing code. One user who cancelled his Claude Max subscription, which costs from $100 a month, uses the assistant for writing applications and tools; he's worried the watermark might appear even when the text is original and Claude was only used for translation, fixing citations, or summarising, threatening problems in academic and professional settings with strict AI rules. He switched to Cursor and Grok 4.6 for coding and Grok Bot for work and personal tasks. Another former subscriber, who used AI for code review and translation, fears the mark in code sent to clients could raise authorship questions and contractual penalties, and he chose Chinese AI providers instead. A third uses Claude for business documentation but worries about marks in texts he originally wrote himself, which the bot only edited. A fourth, a software engineer who cancelled his Claude Max subscription back in July, cited declining service quality and limited access to the best models, and said the watermark incident only confirmed his decision, as he didn't want professionally produced code to raise authorship or policy questions. Positive aspects of watermarks include transparency about content origin, as more content is AI-generated, and protection against one AI model training on another's materials. Besides Anthropic, Google uses its SynthID technology, and Elon Musk's X AI assistant also watermarks content, which recently briefly appeared in an official White House statement. Anthropic says it added the watermark to comply with new EU regulations and promised to release a free API for checking content for Claude watermarks. One now-former client said the initiative intensified his concerns about over-reliance on Claude, noting that closed software can be changed at any time by the developer, so the AI used in work shouldn't be vendor-dependent.