# Anthropic's New Watermarking Policy Sparks User Backlash Invisible markers in Claude's outputs will detect AI usage in professional and academic settings. - **Source:** [TechCrunch](https://techcrunch.com) - **Published:** 2026-08-12 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/8ea3669c-8d61-4896-b8fc-4622741d6f58 ## What is Watermarking? Anthropic has started embedding invisible digital watermarks into Claude AI's text outputs. These watermarks allow computer systems to detect whether text was generated by artificial intelligence. Implemented to comply with the EU AI Act's Transparency Code, this practice has sparked controversy among users who fear increased surveillance of their AI usage patterns. ## EU Regulation Mandates Transparency The European Union's AI Act, effective from 2026, requires technology companies to label content created or edited by AI. Anthropic's watermarking fulfills this legal requirement. Critics argue the implementation creates unnecessary monitoring, while supporters maintain it's essential for maintaining transparency in an increasingly AI-driven content landscape. ## User Reactions Divided On Reddit, some users describe the watermarking system as "draconian," particularly students and professionals concerned about being caught using AI. One commenter warned, "The student who used Claude to reorganize a paragraph will be flagged." However, other users counter that objections mainly come from those wanting to conceal AI assistance in their work. ## Ethical Debates Intensify The controversy centers on ethical questions about AI use. Copying AI outputs verbatim already violates academic and professional ethics. Supporters argue watermarks will help prevent such violations. A Reddit user stated, "The only reason to oppose this is if you want to lie about AI use," highlighting the transparency argument driving the policy. ## Irony in Implementation Some critics find irony in AI models watermarking outputs after training on others' work. One user commented, "Given how frontier models obtained training data, this is deeply cynical." This perspective reignites debates about data ethics in AI development, questioning whether companies should claim ownership over generated content. ## Impact on Professional Workflows Journalists, writers, and developers who regularly use AI tools may need to adjust their workflows. For instance, a journalist using AI to summarize a 200-page transcript would need to properly edit rather than directly paste watermarked output. The policy could significantly impact productivity in content-heavy professions that have come to rely on AI assistance. ## Attempts to Circumvent Some users plan to bypass watermarks by reprocessing outputs through different AI services or manual editing. However, these methods may prove time-consuming and impractical. Experts predict watermarking technology will become more sophisticated, making circumvention increasingly difficult as detection methods improve. ## New Era in AI Usage Anthropic's watermarking may herald a new phase in AI adoption, with other major companies likely to follow suit. This development represents a significant step toward responsible AI use and content verification. While controversial now, such measures may become standard as society adapts to AI's growing role in content creation. ## Sources covering this story - https://techcrunch.com/2026/08/12/some-claude-users-are-mad-that-anthropics-new-watermarks-will-catch-them-cheating-at-their-jobs-classes/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._