60. The Hidden Risks of AI: What Every Therapist Needs to Know

By /

The Hidden Risks of AI: What Every Therapist Needs to Know

Are you aware of the hidden risks of AI in your therapy practice? In this episode of Run Your Private Practice with AI, I’m unpacking some eye-opening truths about how AI tools like ChatGPT may compromise your client’s privacy. From recent statements by OpenAI’s Sam Altman to the rise of AI therapy pods, I’m sharing real concerns and practical takeaways to help you stay informed and ethical while using AI.

AI Tips You Don’t Want to Miss:

  • Why anything entered into ChatGPT can be subpoenaed
  • The privacy pitfalls of de-identified client data in AI tools
  • What to know about AI therapy pods popping up in public places
  • How AI agents might be accessing more than you realize
  • Why secure, HIPAA-compliant AI tools are a must

Mentioned in this episode:

Connect with me:

Sponsored by Alma: www.helloalma.com/kym
Sponsored by Berries: www.heyberries.com/therapists

Transcript:

[00:00:00] Hey guys, I am just sort of checking in, every once in a while, I like to just do an update here around just trends I’m seeing with AI trends I think maybe we should be aware of or paying attention to.

[00:00:13] So, Sam Altman of OpenAI. You know the founder of ChatGPT, the one that runs ChatGPT, he did a live video basically saying that anything you put in ChatGPT is something that could be subpoenaed, that your records could be subpoenaed.

[00:00:29] And he was going on talking about basically people put their deepest, darkest thoughts into ChatGPT thinking it’s anonymous and you know, it’s not, he was like, we literally, if there’s a court order, we have to give you record. So he was talking all about that, and it got me thinking of course, it’s like, you know, last week or the week before I did one of these lives and I just talked about like ways to handle when your client brings AI into the room.

[00:00:54] This is a whole new level that you can talk to your clients about, about your records can be subpoenaed, so just reminding them to be super careful about what they’re putting in and us too, right? I mean, we, obviously, I’ve said a thousand times in this group and on this podcast that you cannot put PHI into your ChatGPT or any Claude.

[00:01:14] you know, Gemini, any of the free AIs that you might be using out there. And a lot of us think like, okay, I’ll just have this chat thread. I won’t use my client’s name. But you know, it’s really easy to get in that trap of just putting a lot of information that if your records were subpoenaed, let’s say, God forbid a client

[00:01:32] committed suicide or something, right? And somebody decided they were going to subpoena your Chat GPT records. I mean, who knows? And you’re using that information inside of, of ChatGPT, and you think you’re like de-identifying the client. But there could be a lot of information in there that when it’s all put together, you know, it’s pretty clear who you’re talking about.

[00:01:50] So you just wanna be super careful. Like one post might be like, oh, they talked about their new cat. The other one might be like their abusive husband. And you think that they’re de-identified, but literally if you paste it all together, it might spell out exactly what was happening with the client.

[00:02:02] So two things are really standing out about that. One is when your clients are bringing AI into the room, into the therapy room, reminding them anything you’re putting in here is something that could go to court and be used against you. Not that most people are putting like criminal activity or anything in like that, but you just, wanna be super careful what you’re putting in.

[00:02:21] And for us too, as therapists. That’s why of course, using a HIPAA compliant note scribe that is behind a security wall is super important like Berries or whatever note scribe you like to use, and you guys know I, I love Berries and I’m the clinical consultant for them. So of course Berries would be my go-to.

[00:02:38] But making sure that you’re doing it that way. So that was a really interesting video that I was watching and I’ll, like I said, I’ll get the link and put it if you guys wanna hear what he says about it. But he’s like, we haven’t figured out a way to get around this yet. Right? We don’t know how to protect your confidentiality with ChatGPT, is what he said.

[00:02:56] So keep that in mind just to anybody using it. The other thing, somebody posted in the Run Your private practice with ai, Facebook group, a AI therapy pod. I don’t know if you guys saw this, but it’s a booth. It’s a prototype. I guess it looked like a prototype from the article. And again, I’ll, I’ll link it below so you guys can see it.

[00:03:14] It’s a prototype, but it’s a booth that’s soundproof. It’s in your workplace or wherever. I mean, maybe they’re gonna put ’em in coffee shops. I don’t know. Walmart, I mean probably Walmart, CVS. I mean, who knows? All of them are like branching out with these therapy type things. It’s a Thera AI pod, and you go in, it’s soundproof and you sit in there and basically an AI therapist avatar comes on the screen.

[00:03:38] You have a quick session with the AI therapist about whatever’s bothering you, and that is your therapy for the day or the week, or if you just need it. Now, some of the comments in the group were saying like. Yeah, I’d never do that, especially in my workplace. Like who knows who’s listening and who can access that data that’s going through, where does the data go?

[00:03:56] Of course, same rule, right? what’s the hipaa? What’s the security? How can people access this information? Is this something that could be subpoenaed at some time? Like you just have to be so careful about the stuff and it’s just. It is pretty crazy. I mean, even me on my phone, right the other day, I don’t know what I was doing,

[00:04:13] some AI option came up. I used it. It was totally personal information about my dad’s medical records. I used the agent the other day, the ChatGPT agent that just got released. I think I might have posted about it in the group. And of course I’ve done a ton with it and I’m waiting to like get a bunch of data before I actually talk about it on the podcast so you guys can know everything I’ve been doing with it.

[00:04:31] Kind of crazy and unbelievable. But even then I had to go out to a website that I allowed it to log into, and I allowed it to access my account that had stored credit card information. It had stored passwords. And somebody in my clinical AI club was even asking, they were like, how secure is this? And I’m like, you know, I really don’t know.

[00:04:51] I got so excited about trying it that I just let it have access to everything, so it’s probably not secure. It did access my account with credit card access information saved. I did allow it to log in using my stored password that was on my computer. So I don’t know. I mean, some hacker gets in, they can access it, but then again, like I feel like I get hacked like three times a year just through like my credit card company or something.

[00:05:15] My passwords get compromised. So I don’t know. I’d love to hear what you guys think about this. Like, would you ever, ever allow access like that with the agent like I did. And, you know, maybe I should get more in depth what I did with it. But I was letting it access my Instacart and order groceries.

[00:05:32] I let it access my Zapier account, which has access to my Outlook email, which has like all my passwords in a note in there, which I know is dumb. But I do that, and I’m saying it out publicly right now, but I don’t know. So. This is just the latest stuff that’s coming up with ai and it just, you know, as people that are using it to run our private practices, it’s important obviously to remember the HIPAA thing.

[00:05:54] But of course, just in general in our personal lives, like how are we using ai? I’ve got AI on my iPhone. I’ve got AI on my all, all my browsers. I’ve got AI in my Word documents. Now Microsoft copilot interacts with my Word documents. My Google Drive. It is connected to ChatGPT I gave it access so it could like, see my brain.

[00:06:13] Like everything I do is in that Google Drive, so it’s got access to everything over there. I mean, one hack and somebody really could have my whole life in the palm of their hands, you know? And I know it’s possible already for a lot of people. A lot of people are hacked all the time and like their lives get ruined because of it.

[00:06:28] it’s just a little bit frightening, the access that I’m giving. That you guys might be giving, that we might unknowingly be giving enough bits and pieces of information to, hurt our clients eventually. So anyway, that’s my 2 cents for today. I’d love to hear what you guys think if you’re using it.

[00:06:46] Have you used the agent? I wanna know more about that. The agent is the possibilities are incredible when you really start playing with it and realizing all it can do, like, I can’t even really wrap my mind around it yet, so I’m still working on that. And of course, I wanna give a, like a fully in depth report.

[00:07:02] The other thing I wanna tell you guys about is in the clinical AI Club this month and the month of August, we have Maria Bed hyphen Diaz. She’s going to be in the group giving, well, she already released it to the Clinical AI Club member, so if you guys are in there, it got released on the 25th. Her ethics.

[00:07:21] Presentation about AI and ethics. She travels the world basically giving this presentation. It’s fantastic. She did it just for us, curtailed to us. And then next Tuesday, whatever that date is, the first Tuesday of the month in August, she is gonna be doing, she’s running and hosting the 90 minute live implementation session that we do every single month in the clinical AI club.

[00:07:40] I do it one month. I have an expert do it the next month. it’s her this month, and she’s. Awesome. You guys are gonna love it. So if you’re in the clinical AI club, make sure you go to that. She’s gonna talk about ethics, but just not even that, just any kind of geeking out thing about AI and private practice and how to stay ethical around it.

[00:07:56] She’s gonna be talking to you guys about that in the implementation session, so I’m really excited about that too. Anyway, that’s all I’ve got for today. Let me know, what do you think about these? AI therapy pods just in random places. I mean, like, what do we do this in airports while we’re waiting for flights?

[00:08:10] I don’t know. It’s cool and it’s scary all at the same time. It really kind of freaks me out in a lot of ways. But of course I’m obsessed with ai, so who knows. I might even try one myself. alright, I’ll talk to you guys later. Let me know in the comments if you have any questions, thoughts.

[00:08:22] . Okay. Talk to you later.