Claude Is Digitally Watermarking Content, and Your CE Trainings Can’t Use AI
A member in my AI club recently froze over a simple question about her CE training slides, right as news broke about Claude watermarking content and rumors started flying about an APA crackdown on AI in continuing education. In this episode of Run Your Private Practice with AI, I break down what the watermarking news actually means, what the APA guidance really says, and the four buckets I use to figure out where AI belongs in the CE process and where it doesn’t.
I also get into why detection tools like Pangram and ZeroGPT are showing up everywhere right now, why asking whether the machine touched your work might be the wrong question entirely, and where I land on the difference between AI generating your ideas and AI arranging the pixels.
AI Tips You Don’t Want to Miss:
- Why LinkedIn and Substack’s new AI detection features are flagging even fully human-written posts.
- The four-bucket framework for deciding where AI belongs in CE prep and where it doesn’t.
- What the APA’s AI disclosure requirement for CE providers actually says.
- How free tools like Pangram and ZeroGPT can check your own content before you publish it.
- Why “did AI touch it” might be the wrong question to ask about your CE materials.
Connect with me:
- Instagram: @thetravelingtherapist_kym
- Join the Clinical AI Club: www.clinicalaiclub.com
Sponsored by Alma: www.helloalma.com/kym
Sponsored by Berries:www.heyberries.com/therapists
Sponsored by Auralink Pro: Start your free trial and use code KYMTOLSON for 20% off www.auralink.com/pro
Transcript:
[00:00:00] Hey everybody. Welcome back to another episode of Run Your Private Practice with AI. If you’re watching this on video, I’m sure I look cute. I just got back from the gym, so ignore that. I just wanna record this episode. I wanna record quite a few episodes ’cause I’m actually getting ready to go out of the country for a sorta like sabbatical.
[00:00:19] I’ll be out of the country for three weeks, so I’m pre-recording a lot of these episodes beforehand, so I don’t have to work. that life of freedom that I hope all of you are living. You know, part of what I do is I have multiple income streams, and my ultimate goal is to be able to just make enough money and have enough systems in place, mainly using AI and my team, that I don’t need to work every single week, which has really been a dream, and it’s a reality now.
[00:00:45] So I’m super excited to be able to go off to Switzerland and then Italy, the Dolomites, with my nephew and my sister and not have to work at all. Sharing that because uh, even today in my Facebook groups, every day somebody says like, ” how do you get more income streams so you don’t have to see all these clients all the time?”
[00:01:03] And that’s how you do it. AI, a team that helps you scale and get out of that one-to-one model. So what I thought I would talk about this week is something that’s coming up everywhere. So Claude recently announced that they’re gonna start putting like a watermark on all of its content that’s generated.
[00:01:23] And this comes out of the UK basically saying that, you know, anybody in the UK using AI has to identify that they’re using it, and there’s now these requirements in the UK that all AI stuff is watermarked. So I guess Claude just decided it would do it for everyone. So as soon as I saw that, in about, I don’t know, maybe 15 minutes, 30 minutes, I was seeing even ads on Facebook for all these people that have created a watermark remover, a Claude water-watermark remover
[00:01:52] So, you know, it’s, it’s just crazy to me, all of it. It’s like, just let us use AI. Like, who cares? If you could tell it’s AI, good. Why do we need to disclose it? I mean, that really is my opinion. I’m sure people listening to this are gonna be like, “No, that’s not how you should be able to do it.” But anyway, so let me just, like, kind of tell you how all this came together and why I wanted to record this episode.
[00:02:12] Somebody in my clinical AI club group asked me a question a couple weeks ago, and I’ve really not been able to stop thinking about it ’cause it’s like, I don’t know, it’s just mind-blowing what’s happening with AI. Every single week something new is coming up. She wanted to know if she’s allowed to use AI to build slides, like a slide deck for her CE training that she was gonna put together.
[00:02:36] Not to write the training, but just to build the actual slides, and she was nervous enough about it that she offered to send me a screenshot from somebody else’s post. I don’t know if you guys know Michelle Riser. She’s, like, kind of known for helping therapists learn how to get CEs for their offers and that sort of thing.
[00:02:53] So my member in the club was reading Michelle’s post, and she was afraid to use AI for anything. She doesn’t want to get in trouble with the CE boards. And basically what Michelle’s post said was that I think the APA was cracking down on you being able to use AI in any capacity to create your CEs, which doesn’t make any sense to me.
[00:03:18] if you search Michelle Reiser, you’ll see more information about it, what she posted in there. But just like sit with that for a second, okay? We’re like licensed therapists. We have real expertise. We have a whole training in our heads that we want to share with other people and also, you know, allow them to get CEs for it.
[00:03:39] And this clinician was afraid to just like open up Canva or, you know, Manus or whatever can do your slide deck for you really easily now. But she was afraid to do that because of these new restrictions that are coming out from the APA. So that’s where we are right now. Now I want to talk about it a little bit today because I think a lot of us are maybe gonna be overcorrecting out of fear with all the stuff coming out.
[00:04:03] Now I get inundated with it, right? I see like… I subscribe to a million AI newsletters, so in the last two weeks I’ve seen a ton of stuff on this like watermarking. But there’s two people I follow that I really do respect, and I want to talk a little bit about what their newsletter said. And then just on my own research, going around trying to do different things to figure out, like, how are we gonna handle this if every single thing we do i- with AI has got like a digital watermark on it now, and, you know, down the road, is it gonna affect our SEO?
[00:04:33] Like I just did a workshop and taught, I mean, there were over 100 therapists. People are emailing me every day with their websites. Like, but we, we built them with Claude, right? Claude wrote the, the c- the copy for it and helped us build it out and created the prompt and the HTML and all that. So is that going to eventually down the road hurt SEO if we’re using these tools?
[00:04:55] You know, who knows? I, I really don’t know what’s gonna happen with it. So anyway, two newsletters, same day, two people I actually respect, Katie Reed, you guys might know her, and Brittany Long. And it was interesting to me. I like to see all these opinions because the two of them came to two totally different conclusions about it. The first one, Katie walked through what’s been happening with AI
[00:05:18] Cut that last part. Sorry, Darren. Cut that So the first email walked through what’s been happening in the world of AI lately, and a lot of it is about detection that’s happening. So LinkedIn rolled out a feature that basically lets people flag your post as looking like AI slop. So if you’re looking at something on LinkedIn now, you could just like flag it, send it to LinkedIn.
[00:05:40] Substack now has a scanner that tells the readers what percentage of your post it thinks was AI-generated. And then of course, there’s the chatter about the watermarking on AI-generated text and
[00:05:53] I don’t– And like I said, I don’t even know what that means legally or anything like that because I don’t think anybody else knows and hasn’t even tried to figure it out yet. But it’s just, it’s coming up everywhere, right? The whole ecosystem is building tools to like catch you using AI, which just I, I just don’t get it.
[00:06:12] It’s like an amazing tool out there to help us be more efficient, proactive proficient, you know, whatever word you wanna use. But we’re like, they’re like trying to catch us. It just doesn’t make sense to me. And, you know, maybe somebody else listening to this, like the, a, a writer or somebody who’s a good copywriter or something might be like, “Thank goodness,” you know?
[00:06:30] “My job’s safe if, if this becomes a real deal.” But to me, it just, I just don’t really understand it. But I am wanting to talk about it so you guys know what’s happening in the world of AI and all of this stuff that’s kinda like converging at once on the same like, like level with LinkedIn and Substack and I’ve seen it in a couple different places too.
[00:06:51] Even on Instagram, right? Now, if you use Instagram’s own edits tool or CapCut or something like that, and you’ve removed the background, now that’s flagged as AI. So it’s like, what can you even do anymore that’s not, that doesn’t really have AI like involved in some part of the process? You know what I mean?
[00:07:10] So we had that first newsletter from Brittany, and then the second one was from Katie Reed, who’s amazing. If you guys don’t know her, she writes “By Behavior.” She’s really into AI and really amazing. She’s an amazing person, knows a ton about AI. Brittany is too. But hers was more of like a personal reckoning, which I thought was interesting, and she really said something that caught my attention in that she says that she writes every single one of her newsletters herself.
[00:07:36] Like her y- her AI newsletters that she’s writing about AI, she writes them herself. And even then, writing them herself, her newsletters are still getting flagged as AI-generated content. So what can you do? Not all of them. She said not all of them were, but the part that she really talked about is the ones that are getting flagged are struct- like structured teaching issues that she puts out, like, “Do this, do this, do this, do this.”
[00:08:03] Those are getting flagged, like step-by-step instructions on how to do things. The ones where she’s explaining a concept, she’s laying out a framework, she’s walking you through steps, that sort of thing are coming back labeled as AI. Her stories that she does don’t get flagged. So if you don’t follow Katie, I’d say follow her.
[00:08:21] She’s really interesting. Her– She’s a fantastic writer, like I look forward to her emails. That’s one of the few that I actually look forward to because she’s a really good writer. But she says when she’s doing that type of writing, that detector kind of goes quiet and it doesn’t pick up that there’s any AI in what she’s doing.
[00:08:37] So her theory is that the machine can’t predict where a, a real story is going, so it doesn’t recognize the pattern. And I saw Katie say that, but I’ve seen quite a lot of people say that, you know, things around like personalizing your content, making sure you’ve got your personal story in it, making sure that you know, you make it like a personalized, l- like life experience that AI is not gonna be able, able to replicate that easily.
[00:09:03] I say that and I see people saying that, but also I’ve had AI write a story for me way back when a few times. Like, “Hey, just make up a story that would sound good to go with this thing I’m trying to do.” And it can totally manufacture a story. So, you know, the more AI knows you, you tell it to tell a story about yourself that it already knows your whole history, right?
[00:09:25] AI’s writing a story about you, but it’s still AI. So the, the line is so gray and blurry for me. But anyway, so
[00:09:32] Anyway, so her theory is that the machine can’t predict where a real story is going, like I was saying before, and it doesn’t recognize the pattern, so then it doesn’t get flagged as AI. But think about what that means for us, like especially if you’re like me, like you’re a therapist but you’re also a coach and you’re putting like content out that you’re trying to sell to the world or just content that you’re trying to use like in blog posts or to be helpful, like teaching AI step by step, right?
[00:09:59] I mean, that’s a very specific thing that you’re doing, and it’s a format that a lot of us write in, right? Psycho-educational stuff, handouts, course mod- modules, CE objectives, blog posts explaining a modality, all of that. That’s the exact shape of the writing that we do a lot of times, and that’s what’s getting falsely flagged by these like AI detectors now.
[00:10:24] So really, you could write every word yourself, right? ‘Cause I mean, I’ve been doing therapy for 30 years. I could tell you step by step how to do a ton of like coping skills you know, conversational skills with people. Like I could just list those out, but they’re gonna look like lists, right? And they’re gonna look like stuff that AI’s able to generate really easily now So anyway, Katie’s response was to pull back.
[00:10:48] She says she’s gonna write her own social media post again instead of letting AI chop up her, her newsletters or anything like that. She’s gonna do, like, go way back. She’s gonna just go, like, real authentic in her writing, which I think is amazing, and she’s a great writer, right? So that’s much easier for her than for somebody like me that is terrible at– I’m terrible at copywriting.
[00:11:09] Like, I can never think of what to say. AI has helped me with that a lot. And she said that, but she also said that it fills her with dread and crankiness, which I really did appreciate her admitting that, that even to her, that’s a difficult thing. She doesn’t really wanna be, like, all in with having to do her own copywriting again for everything.
[00:11:28] In the other newsletter, Brittany’s l- newsletter said almost the opposite, which I thought was interesting. Keep using AI, just don’t start there. So have AI research what your audience is talking about, and then record your own thoughts on it, which is what I do a lot. Like, I walk around with earbuds in when I’m, like, exercising or something like that.
[00:11:47] I’m constantly talking to my AI, right? I’m giving it, like, my thoughts, my feedback, all of that. It goes into, like, Claude, into a main memory database, which, you know, then basically curates Claude to be a version of me, right? So, you know, Brittany’s talking about brain dumping. That’s what I do. But then it gets into this thing.
[00:12:09] It’s like, okay, if Claude is like a mini me because I’ve trained it so much, or ChatGPT, even my ChatGPT knows me really well is it AI-generated content, or is it AI helping me generate the stuff in my brain that I’ve already, like, trained it on? You know what I mean? I think it’s such a gray area there that the stuff just, it just is kind of like a mind fuck.
[00:12:34] You know what I mean? Not to be kind of crass, but it really is when you start thinking about it.
[00:12:39] So anyway, two really smart people that I respect, I think are amazing with two different opinions. One says she’s gonna remove AI from her workflow as much as she can, and the other one is just putting a rule on how it enters into the AI. Me, I mean, I’m just like, I feel like Claude is me. You know what I mean?
[00:12:57] It’s got a nice little brain behind it, but I feel like it’s so curated around all my projects and my thoughts and my .md files and the skills I’ve built and, you know, writing samples and all of that. I mean, I feel like it’s already kind of me. Somebody else went into my Claude, they would not get good output because it’s part of me at this point, which is a little scary to say, but it is how I kind of feel about it.
[00:13:20] So anyway, so everybody is drawing this line, and I don’t know if it’s in the wrong place or the right place, but it’s just interesting how this is coming out right now
[00:13:28] And we’re all kind of like arguing about whether the machine touched the work or not. And to me, that’s really just the wrong question. The question has no bottom to it. As I was saying before, it’s like, did the machine touch it? Grammarly is a machine, right? Spell check is a machine in your Word doc. The autocorrect on your phone is a AI machine essentially.
[00:13:51] Canva’s tools where they’re removing backgrounds is an AI feature, right? All of this is AI. So it’s like if a machine touched it, is your standard like, “I’m never gonna have a machine touch anything that I do”? How are we gonna do that? It, it doesn’t even seem possible to me.
[00:14:08] It’s like you have to go back 40 years into the past to be able to you know, create like original content that AI hasn’t touched, in my opinion. I mean, even when I type a post in Facebook, Grammarly pops up, it corrects my grammar, I click three buttons. Sometimes it’ll make a suggestion about how to phrase it better.
[00:14:26] I’ll, I’ll say, “Oh, that does sound better.” I’ll click it. That’s AI, right? If all that stuff starts getting flagged, like it’s just basically the whole world’s gonna be flagged, in my opinion. You know, I’d love to hear what you guys think about this too. If you’re listening, hit me up, send me an email at info@kimtolson.com, because there’s just so many different sides to it.
[00:14:45] And then we’ve got the people that are like, “I would never use AI,” but they use AI in everything in our lives, right? We watch Netflix. Netflix like is an AI algorithm, right? It shows you something completely different than it shows me based on my interests. You know, I could just go on and on about all the ways AI is just working in our life every single minute of the day, really.
[00:15:06] I’m recording this on something called Riverside. As soon as I’m done, it’s gonna give me the option to to AI-ify and take out silences and filler words and, you know, any retakes I did, I can like hit that up and like if I said something twice, it’ll take out the double words in there. That’s AI, you know?
[00:15:25] But it makes my life a lot easier to use a tool like this to make this process faster for me so I can put this content out, right?
[00:15:32] So anyway, going back to like Michelle Riser and the CE training stuff I’ll, I’ll see if I can find her post again. But it’s interesting. Go, go look her up. You know, interesting that this is starting to come down from, I think it was the APA. I should probably go look and check that. I think, I’m pretty sure it was the APA.
[00:15:48] Let me go look. I wanna make sure I’m giving you guys the right information. Let’s see
[00:15:53] Okay, the APA now requires CE provider applicants to disclose any AI used in the application process. They do not allow you to use it at all in actual course planning or creation, is the words that she used, okay? So creation, that was where, you know, my member of the club was talking about even slides, using it to create slides.
[00:16:14] Like what does creation mean? You know? This to me is just, just insanity really. Okay, so
[00:16:22] This is how I think about it, and this is what I told my member, just, you know, this is my thoughts on it. Bucket, like different bucket, like four buckets, right? To think about this in. So bucket one is like substance, and the substance is yours, right? It’s coming out of your brain. It’s your outline, it’s your talking points, your learning objectives, what you think is true about a topic, what you think matters, what you decided to teach, and what you, and what you decided to leave out of the training.
[00:16:49] That all comes from your brain, like your clinical judgment based on the topic that you’re teaching about. That’s your intellectual property and your clinical voice, and it should stay non-negotiable, right? You don’t want to go to AI and say, “Write this whole thing,” although I have done that before just to see if it would do it.
[00:17:07] But go over and say, you know, “Write, write a whole presentation about this topic,” and then just use it, right? You obviously want it to come from your experience and what you know about the topic. Yeah. It’s like if you outsource that layer, it, it– you have nothing of it that’s even yours. It’s just, I mean, not AI slop, but it’s just like regurgitated from a machine. So you’ve gotta make it yours, right? The reason that people register for your training is that you’re the one who figured something out, or you’re the one that has the modality or the protocol or has the lived experience that knows exactly why this thing works the way it does and why it’s important to teach you about it.
[00:17:45] And if you give that away, I mean, that’s, in my opinion, like what you shouldn’t do. I know people do it, but that’s, in my opinion, what you shouldn’t do
[00:17:53] Then there’s like bucket two is a structured like feedback, and I think that one’s fine too. So once I write my own outlines about things, I’ll paste it into Claude and ask something like, “Does this outline flow logically for, you know, a 60-minute presentation? Does it drag somewhere? Does it need more explanation here?
[00:18:13] If you were like listening to this for the first time, do you think you would be able to learn these objectives from what I put in here?” Right? That kind of thing. Or saying like, “Based on this content here, can you pull out the objectives that meet the requirements for the CEs?” That kind of thing. Lisa, I don’t know if you guys know Lisa Mustard, she is the “That Therapy Show” podcast.
[00:18:35] She also has podcourses with CEs, but she’s got even a GPT that she made for people for free that helps you pull out like learning objectives and that sort, sort of thing to make the CE process easier. She also teaches about how to get CEs, and she helps facilitate that too. So that’s like feedback, in my opinion.
[00:18:55] That’s the same thing as if I sat down with like a friend of mine that also specializes in this thing and said, “Can you go over this with me? Give me some feedback. Is there anything you would change?” That kind of thing. And really, I mean, I don’t know. Is getting a colleague’s opinion on your CE presentation not okay?
[00:19:10] Is that cheating? I don’t think so, but maybe, maybe they would think so. My opinion, it’s not, right? It’s just like running it by somebody, and I do that with AI all the time. So bucket three is like the polish on it, right? Which I also think is fine. So like tightening a sentence that might be too long or maybe a little like less clear in what I’m trying to say.
[00:19:34] Fixing the fact that I wrote the same word four times in one paragraph. You know, it’s like, oh, there could be a little more variety in this, or you already kind of said that right here. Or like sometimes I’ll like tell it to edit it and make the bullet points that I have like actually fit on the slide, right?
[00:19:49] Like, “Hey, can you like condense this so I can put it on the slide more like succinctly?” That kind of thing. To me, that’s fine, right? I mean, I’m not the APA, obviously. I’m not like approving CEs, but to me, I just don’t understand how that would be an issue. And I’m saying all this here so you can think it through yourself if you’re ever gonna be doing something like this
[00:20:08] And then there’s like bucket four, right? Which is like the design piece. Like how many of us are graphic designers, right? None of us. And this is one that had my group member really frozen. It’s like, “Oh my God, now you’re telling me I gotta go like make my own slides?” Like holy crap. Because you know, Canva, Gamma Manus Claude can even do some slides now.
[00:20:29] I don’t think they’re the most amazing, but Claude Design can do slides. Nana Banana does great slides, you know, and it’s so easy. You just like drop in your outline and it’ll turn into these really great slides. Manus, I use Manus for that all the time. It’s excellent
[00:20:42] And how is that different than like hiring a designer to do it for you, right? It’s the same thing. So are they gonna say you can’t hire designers to make your slides? Like, just doesn’t make any sense to me
[00:20:53] So anyway, the rules as I read it, and this is to my member in my membership, my AI membership, and to anybody that’s maybe seen this stuff float around, like it’s about who generates the ideas, not who arranged the, like, like the pixels. You know what I mean? So that’s, that’s my take on it. So, the, the next, like, part about this I wanna talk about is like the test.
[00:21:19] If you’re gonna do a test or you wanna do a test about it. Katie shared in her y- newsletter some great resources for this, and there’s one that I’ve been using too that she shared that you could take your AI, your, your AI-generated, your real generated text or whatever, you could plug it in over there, and it comes back and gives you a report and says, “This is not AI-generated content.”
[00:21:41] Now, I think it’s hilarious because I have like a social media content creation. Those of you in the Thera AI Hub have access to it too. It writes like social media for me and that stuff. I was like, “Well, let me see what it says,” you know? And so I took, pulled out some of the, the captions for my Instagrams or whatever, plugged it in, and it was like, “This is 100% human-generated.”
[00:22:03] I’m like, “Okay. It’s definitely not a h- 100% human-generated.” Now, I do use a skill that I created called Kim Anti-Slop, Anti-AI Slop skill that I created that I will turn on and kinda hit it and say, “Hit all of these with the Anti-Slop, Anti-AI Slop j- like skill to go over,” and it’ll kind of rewrite some stuff now.
[00:22:25] And I did use it from there, but it’s still generated by Claude. I copied it from Claude, put it into the thing, and that’s what it said, you know?
[00:22:35] So the tool that Katie talked about, the first one was P- Pangram, P-A-N-G-R-A-M. That’s actually the one that they are using in Substack. I saw like a press release recently about Substack was gonna be allowing users to use this, and I go to Substack a lot, and I have seen it in there. So Pangram. And then the one I use and that Katie talked about too is ZeroGPT, and it’s free.
[00:22:57] So you can just open it up on your browser. I think the, the the trying to use it on your phone is a little bit tough, but if you open it up on your computer browser and just plug in something, you’ll get a report back basically that tells you if it’s AI generated or human generated.
[00:23:11] So anyway, I guess if you’re gonna take anything out of all of this, it’s just that there’s a new– it’s like a new dawn with AI stuff, right? It– there’s gonna be detectors everywhere. And I really predict that we’re just all gonna ignore the detectors over time because it’s gonna be everywhere, right? And it’s like there’s gonna be people saying, “I wrote this myself,” and other people saying, “No, AI wrote this.”
[00:23:36] You know, so I kind of predict that, like, even if everything’s watermarked, one, as soon as the watermarks come out on Claude, a day later there’s like 15 watermark remover tools out there that you could use if you wanted to. There’s tools like Pangram and ZeroGPT that you could use to correct the stuff.
[00:23:57] But really, I think really coming at it from, “Let me just give it my core ideas,” kind of like Brittany said, “Here’s my core ideas, and then help me polish it,” I think is okay, you know? I’ve seen people saying that, “Okay, I’m just gonna start dic- disclosing in everything I write,” like, “AI was used at some part of this process.”
[00:24:15] But imagine, like, if you were trying to get CEs or something, right? And you had to, like, report back to the CE person every single piece that AI touched, like if Grammarly, like, edited your notes or your slides or something like that. Like, could you imagine if you had to, like, at– really list all that stuff?
[00:24:37] So I don’t know. I think it’s overboard. I think we’re in an interesting world right now. Who knows what’s gonna happen with all of this? I just wanted to bring your awareness to it and really tell you about the tools. So the ZeroGPT is one that I’ve been using for a little bit to kind of check things just to get a feel for you know, if it seems like AI, because I think I’m in it so much, sometimes I don’t even realize it sounds AI-ish, you know?
[00:25:04] Even though I’ve got all these tools to, like, make it more humanized, that kind of thing. So anyway, that’s what’s going on with AI and the Claude watermark and, you know, these things in Substack and LinkedIn and all of that. So I’d love to hear what you think about it. Am I way off base with this, thinking that, you know, if it’s our original ideas and AI helps us, is it AI’s or is it ours?
[00:25:29] That’s kind of where I’m with this right now. Okay. Talk to you guys later. Have a great day. Bye-bye.