SPECIAL SERIES ==> 🚨 Your AI Copy Is Snitching On You <== | BATHROOM Break #121 COLLAB: The Marketing Millennials + Do This, Not That
August 17, 2026
AI Summary
5 min readAI Watermarks Are Coming for Your Copy—And You Probably Didn't Notice
Jay Schwedelson and Daniel Murray are sitting in a New Orleans hotel room, Jay still dusted with beignet powdered sugar after moving his son into college, when they drop a warning that most marketers aren't ready for. Both Claude and ChatGPT have quietly started embedding watermarks into their outputs—not just in images, but in text itself. And this changes how you can safely use generative AI going forward.
Claude's Text Watermark Is Already Live
The first major development: Anthropic's Claude is now inserting an invisible watermark directly into the written text it produces. This isn't metadata hidden in a file header. It's structural—woven into how the sentences are constructed. If you copy and paste text straight from Claude into another platform, detection tools can now flag it as AI-written.
Jay frames it bluntly: "The age of copy and paste from Claude, specifically text or images, is over." The watermarking started with European Union AI transparency requirements, but Anthropic rolled it out worldwide. It's happening right now.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Do This, NOT That: Marketing Tips with Jay Schwedelson
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 SPECIAL SERIES ==> 🚨 Your AI Copy Is Snitching On You <== | BATHROOM Break #121
- 2 (00:07) **Empty Nester Banter & Travel Stories** - Jay and Daniel catch up on Jay's kids leaving for college and travel mishaps
- 3 (01:53) **The AI Watch Out Era Begins** - Introduction to the episode's main topic: major changes in how AI tools are marking their outputs
- 4 (02:25) **Claude's New Watermarking System** - Anthropic is adding watermarks to both images AND text outputs
- 5 (04:01) **The First Draft Rule for Claude** - Practical advice on how to use Claude going forward
- 6 (04:28) **ChatGPT's Image Problem** - Similar issues are bleeding over to OpenAI's platform
- 7 (05:30) **How Far Does the Watermarking Go?** - Daniel questions whether simple grammar edits get flagged
+ Full timestamped outline available in the app
Guests on this episode
Show Notes
Partner with Jay: https://www.jayschwedelson.com/contact
ã…¤
Check out Jay Schwedelson's new book: Stupider People Have Done It
All net proceeds are donated to The V Foundation for Cancer Research, let's kick cancer's butt: https://www.amazon.com/Stupider-People-Have-Done-Marketing/dp/1637635206
ã…¤
Subscribe to Jay's newsletter for weekly marketing tips and tactics: https://www.jayschwedelson.com/newsletter
ã…¤
Register for GuruConference (FREE + VIRTUAL!) https://www.guruconference.com
ã…¤
Check out Eventastic (FREE + VIRTUAL!) https://www.eventastic.com
ã…¤
Connect with Jay on LinkedIn: https://www.linkedin.com/in/schwedelson/
Check out Jay's YouTube channel: https://www.youtube.com/@schwedelson
Check out Jay's Instagram: https://www.instagram.com/jayschwedelson/
Ask Jay anything: https://www.jayschwedelson.com/ask
ã…¤
Leave a comment and follow the show, it really helps us out!
ã…¤
Follow Daniel on LinkedIn and check out The Marketing Millennials podcast for sharp, no-fluff marketing insights. Subscribe to Ari Murray’s newsletter at gotomillions.co for sharp, actionable marketing insights.
ã…¤
Copy, paste, post has been the default AI workflow for a lot of marketers, and it quietly stopped being safe. Jay Schwedelson and Daniel Murray dig into the watermarking wave hitting Claude and ChatGPT, why it now lives inside the writing itself rather than the metadata, and what still slips past detection right now. There is also a beignet incident and a boarding gate betray
More from this podcast
Do This, NOT That: Marketing Tips with Jay Schwedelson →