Claude Max subscribers cancel over Anthropic's new watermark

Four paying Claude Max subscribers told Business Insider they canceled after Anthropic started embedding an invisible watermark in Claude's output. Their complaint is not that the mark exists, it is that the mark sticks even when Claude only proofreads, translates, or lightly summarizes their own writing.

The watermark is Anthropic's move to comply with the EU AI Act, and it is being applied globally rather than only in Europe. According to TechCrunch, it applies to text from Claude models launched on or after August 2, 2026, with older models to be covered by December 2. Generated text carries the invisible mark and it travels through copy and paste; generated .svg, .png, and .jpg files instead carry signed provenance metadata under the C2PA standard. Anthropic's own line is that a detected mark indicates Claude may have processed the content, but it does not confirm Claude authored it.

The users Business Insider quoted sit in exactly the workflows where that distinction gets expensive. Arturo Villarroya canceled his Max plan Wednesday, worried the mark could stick when Claude only translated text, cleaned up grammar, or gave his original writing a light summary. Vladislav Rajtmajer, a Czech Republic-based freelance developer, canceled Tuesday over the risk that an AI label on client code could raise authorship questions or trigger contract penalties. Richard Echols put it flatly: "Even if you edit your own work that you wrote, they add the watermark anyway." Stevan Bogosavljević, a senior software engineer based in Montenegro, had already left in July over service quality and said the watermark confirmed the decision.

Three experts in our Who's Who directory circulated the piece, which sits alongside coverage from outlets including Gizmodo tracking the same feature. Anthropic itself told reporters it has not yet seen a statistically significant uptick in cancellations, so four named users and dozens of frustrated X posts are not the same as a mass exodus, and Business Insider never puts a number on actual Max churn.

The bet Anthropic is making is that regulatory trust and provenance are worth more to the company than the edit-heavy power user who blends Claude output with their own writing. If competitors do not follow, editing and translation quietly become the workflow that migrates first, and the promised binary AI-detection tool will need a false-positive rate the working freelancer can actually live with.