TLDR
- Anthropic will embed invisible watermarks in all text and files created by Claude AI models worldwide
- The move brings Anthropic into compliance with the EU’s Code of Practice on Transparency of AI-Generated Content
- Watermarks will travel with copied and pasted text and use the C2PA open standard for files
- Users cannot opt out, and the policy applies to Claude, Claude Code, Claude Cowork, and Claude Tag
- Backlash erupted on X and Reddit, with users raising concerns about proofreading, coding output, and AI detection
Anthropic confirmed it will begin watermarking all text and files generated by its Claude AI models. The policy, announced August 11, applies globally and covers all Claude products, with no opt-out option for users.
🚨HUGE: Claude will now invisibly watermark all AI-generated text, making it detectable even after being copied and pasted.
The watermark tagging will apply to all new Claude models, including API outputs, worldwide with no opt-out option. pic.twitter.com/YVQjIulimR
— Coin Bureau (@coinbureau) August 12, 2026
The watermark is embedded directly into generated text, meaning it travels with the content when copied and pasted elsewhere. For files such as images, Anthropic is using the C2PA open standard to embed digitally signed metadata.
The change is driven by the EU’s Code of Practice on Transparency of AI-Generated Content, which took effect August 2. Nearly 200 companies, including Meta, Microsoft, and OpenAI, have signed on to comply.
How the Watermarks Work
Anthropic says the watermarks will be applied at the model level, so they appear regardless of which Claude product or surface is used. This includes the Claude platform API, Claude Code, Claude Cowork, and Claude Tag.
The company has warned the watermarks are not foolproof. Heavy editing may erase a text watermark. File metadata can disappear if a file format changes, a screenshot is taken, or the file is converted.
Anthropic says it will publish more details on how to detect the watermarks but has not yet clarified how much editing is needed to remove one.
User Backlash
The announcement triggered immediate criticism on X and Reddit. Users called the policy “hugely problematic” and said it signals the end of “unprovable AI writing.”
Many of the complaints came from people who use Claude solely for proofreading their own work. Radio host Erick Erickson wrote on X that he had switched from Grammarly to Claude for proofreading, and now feels his own writing will be falsely flagged as AI-generated.
Coders also raised concerns, warning that a cryptographic signature embedded in code could degrade output quality.
The announcement also sparked a broader debate about credit for AI-assisted work. Some argued that the person giving instructions should own the content. Others said the watermark is simply disclosure, not a claim of authorship by the AI.
Industry Context
Anthropic is not alone in moving toward watermarking. Google has watermarked AI-generated images since 2023 and has since expanded to text, audio, and video. OpenAI uses invisible watermarks on images but has held back on text watermarking, citing concerns about false positives and driving users to rivals.
AI music platform Suno also announced watermarking last week following legal challenges. Newsletter platform Substack partnered with Pangram to flag AI-generated content last month.
Anthropic has said the watermark policy will extend to older Claude models over time.
Stop guessing and start investing with confidence. KnockoutStocks gives you the AI insights, market intelligence, and stock research you need to spot opportunities, cut through the noise, and make smarter investment decisions — all in one powerful platform.
Sign up today and get 50% OFF full access to our premium stock picks.
Simply use coupon code SPECIAL50 at checkout to claim your exclusive discount.







