The Edward Show
The Edward Show
Edward Sturm
Claude Is Watermarking Everything It Writes. Google Doesn't Have the Key
15 minutes Posted Aug 18, 2026 at 1:46 pm.
Claude Watermark Panic 00:31 What Watermarking Means 01:58 EU AI Act Compliance 03:08 How Text Watermarks Work 03:58 Limits and Detection Keys 04:25 Google Rankings Impact 06:07 SEO Community Reactions 07:22 My Skeptical Take 08:49 Real Risk Reputation 11:23 AI LoFi Success Story 13:24 Using AI for SEO 14:54 Wrap Up and Thanks The Edward Show. Your daily search engine optimization podcast: https://edwardsturm.com/the-edward-show/  #searchengineoptimization #seo #digitalmarketing #seopodcast
0:00
15:18
Download MP3
Show notes
E1140: Claude is now watermarking everything it writes, and the SEO community is worried about what this means for AI content and Google rankings. But there's an important detail: Google likely doesn't have the key needed to verify Claude's watermark. I break down what Anthropic's new watermarking system does, why it's being implemented, what it can and can't prove, and whether any of this should change how SEOs use AI-generated content. Anthropic is adding machine-readable marks to Claude's output as part of its compliance with the EU AI Act. The watermark can remain in text that has been copied and pasted, and Claude-generated files can include provenance metadata. The system is being rolled out across Claude, including its API and third-party platforms. But a watermark doesn't necessarily mean Claude wrote something from scratch. If you give Claude your own writing and ask it to fix grammar, translate it, summarize it, or otherwise process it, the resulting text may still contain the mark. And if no watermark is detected, that doesn't prove Claude wasn't involved. In this episode: - Why Anthropic is watermarking Claude's output - How AI text watermarking works - Why the EU AI Act matters here - Why switching from Claude to another major AI model won't solve anything - What Claude's watermark can and cannot tell you - Why short passages and heavily edited text are harder to verify - Why Anthropic's secret detection key matters - Why Google currently can't validate Claude's watermark - Whether Google could use AI watermarking as an SEO ranking signal - What Ahrefs' research says about AI-generated pages ranking in Google - Why AI content and bad content shouldn't be treated as the same thing - Where AI watermarking could matter for agencies, clients, and contracts - The potential EU disclosure requirements for publishers - Why human review and editorial responsibility matter - My take on whether this changes anything for SEO - Why generic AI content could still create problems even without an AI-content penalty - Why trust, original data, strong opinions, fact-checking, and editorial review matter more than whether AI touched the content - An example of AI-generated content that I genuinely enjoyed - How we're using AI-assisted SEO content on a new SaaS while targeting bottom-of-funnel keywords I also discuss Ahrefs' data on AI content in Google. Their research found AI-generated and heavily AI-assisted pages ranking throughout the search results, including at the top. The bigger issue appears to be something SEO has dealt with long before generative AI existed: bad content. My view is simple: if the output is enjoyable, novel, useful, accurate, and properly reviewed, the fact that AI was involved doesn't automatically make it bad content. The bigger long-term question is what happens to brands that publish large amounts of generic AI content without developing a recognizable voice, adding original information, taking meaningful positions, or earning trust with their audience. AI can help make great content. It can also make a lot of mediocre content very quickly. The difference is what the person using it actually puts into the process. ⭐️ Ahrefs: Claude Now Watermarks Everything It Writes - https://ahrefs.com/blog/claude-now-watermarks-everything-it-writes/  ⭐️ My post on 𝕏 - https://x.com/edwardeachday/status/2089307227064500665  ⭐️ Ep 1110 - Google's Next SEO Apocalypse: The Scaled AI Content Crackdown Is Coming - https://www.youtube.com/watch?v=0w66ME9-IRQ  ⭐️ Ep 1124 - YouTube's AI Detector Is Punishing Real Creators - Is Google Search Next? - https://www.youtube.com/watch?v=33QKtsBy9As  ⭐️ Super Mario Bros 3 Jazz Lo-fi Remix - https://www.youtube.com/watch?v=KNjuiOd1c1I  💎 Compact Keywords - My SEO Course - Get paying customers through SEO - Clear step-by-step video breakdowns - SEO templates to be copied and adapted for your products and services: https://compactkeywords.com/  00:00 Claude Watermark Panic 00:31 What Watermarking Means 01:58 EU AI Act Compliance 03:08 How Text Watermarks Work 03:58 Limits and Detection Keys 04:25 Google Rankings Impact 06:07 SEO Community Reactions 07:22 My Skeptical Take 08:49 Real Risk Reputation 11:23 AI LoFi Success Story 13:24 Using AI for SEO 14:54 Wrap Up and Thanks The Edward Show. Your daily search engine optimization podcast: https://edwardsturm.com/the-edward-show/  #searchengineoptimization #seo #digitalmarketing #seopodcast