Ep.37 | The dangers of sharing medical information with AI (Guest: Linn Freedman, Chair Privacy & Cybersecurity & AI Teams at Robinson + Cole)

Show Transcript

[00:00:00] Hello, and welcome to episode 37 of Privacy Lawls, where I, Donata Stroink-Skillrud, speak with amazing privacy professionals, and we have some laughs along the way as well. Today, I’ll be speaking with Linn Freedman about the dangers of sharing medical information with AI. Nationally ranked in Chambers USA in Privacy and Data Security since 2012, and globally since 2019, Linn is a leader in the field of privacy and cybersecurity.

Linn is a partner at Robinson and Cole, and is the chair of the firm’s Data Privacy and Cybersecurity practice, and co-chair of the Artificial Intelligence team. She has a particular focus on health information technology, and has assisted clients with navigating laws concerning the access, use, and disclosure of protected health information, and substance use disorder information, and HIPAA compliance.

Linn, thank you so much for, for joining me on the podcast today. Sure. It’s my pleasure. Awesome. So really, [00:01:00] people put anything into AI nowadays, you know? It’s from photos, to emails, to relationship issues, to recipes, and now to their medical issues. So when it comes to medical information, what types of medical information do people input into these AI tools?

Is it test results, symptoms, prescriptions, disease history, or, or something else? So, um, our understanding is that people are putting everything in. People are putting all sorts of their very highly sensitive health information into generative AI tools, including, you know, full medical records, um, as- you know, i- inputting diagnoses, prescriptions, and, uh, including some of their personal information that might be attached to medical information.

So really, um, it’s the, it’s the full, it, it’s the full scope. [00:02:00] Yeah. So when they input this information, what kind of questions are they asking the AI regarding their health? Uh, in, in many instances, it’s asking for, um, a summary or really layman’s terms around some information that they see in their medical record.

There might be diagnoses, there might be test results, and they don’t understand what the medical record is saying because they’re not medical professionals. So instead of, you know, calling the doctor’s office or going into the portal and asking questions, everybody has access to their medical records online.

And so, um, a lot of times it’s just to interpret difficult medical information, uh, i- into terms that they can understand. That’s a lot of it. [00:03:00] In addition, uh- We are understanding from medical professionals, and anecdotally we’re hearing, that, uh, people are putting in symptoms into a generative AI tool and asking what they have.

Mm-hmm. So they’re, they’re, uh… If, if they aren’t satisfied with what their physician is telling them, or they’re not, uh, understanding, or they think maybe that it’s wrong, they’re second-guessing medical professionals, uh, frequently. Yeah. I could see that being very frustrating for, for a doctor because, you know, we’ve had Google for, for quite some time, and people use Google to research their symptoms or different medications, and now it’s just so much more accessible to be able to, to question your doctor or what they’re [00:04:00] telling you.

And I know from talking to doctors that they’re very frustrated by this Google armchair doctor experience, and this seems kind of similar. It’s very similar, um, but it’s, it’s probably a, a little worse because at least with Google, it’s, it’s using, you know, it’s using sources. WebMD, the same thing. It’s using medical sources to usually, or researched articles to respond to a particular question.

When it comes to… A- and usually, frankly, the responses that you get from Google or another search engine or, um, WebMD i- it’s pretty generalized and they always say, “Ask your physician,” or, “Go,” you know, “Confirm this with your physician.” With regard to a generative AI tool, um, there, there’s a [00:05:00] real risk that the individual who’s inputting questions around symptoms and, uh, requests for information on diagnoses may get false results, what we call hallucinations.

And, uh, the risk of that is, is higher with generative AI than it is with our traditional search engines or, you know, going to a, a big hospital system and putting symptoms in, um, where, or, or getting more information about a diagnosis. So it’s not as reliable, and it could actually produce, uh, false information that, um, could be relied upon by the individual to, uh, really significant consequences.

Yeah, so this, this brings us really well into our next question. You know, we’ll talk about the privacy implications of this, [00:06:00] but what are the, the health implications of this? You know, um, could AI give you advice that poses a threat to your health? Uh, for sure. And we already know of multiple lawsuits that have been filed against generative AI tools that pr- produced what ultimately was false information to the individual that they relied upon to their detriment.

And in fact, in one case, uh, uh, uh, a, a gentleman basically said, “You know, it, it almost killed me.” Wow. And so I do think that, uh, that people need to understand that they can’t rely on the response given to such an important topic of a diagnosis and in, and in particular a treatment. You, you should not be relying [00:07:00] on generative AI to tell you how to treat something.

You should go to a physician and, and, um, actually have a medical professional, uh, diagnose and treat you. I think that generative AI is very helpful in, uh, navigating questions, but I think it’s a real slippery slope when it comes to, um, relying on it for an actual diagnosis and, and, and, and even more so in treatment.

Yeah. It’s not a doctor. It’s… And, and it’s got, it, it’s scraping false information off the internet. So, uh, I think that, uh, it, there’s, there should be a lot of caution there I, I totally agree. I like to use this example in my work where, you know, I’m not an electrician, so I’m not gonna mess around with the sockets.

And it’s very similar with AI and medical information. Like, if I [00:08:00] tell it, you know, like, “Look, I’ve got poison ivy. I know that’s what this is. I’m looking at this cream that has these ingredients. What’s a Walgreens alternative to this cream that has the same exact ingredients?” It tells me, I can check that.

But it, it’s not good for like, “I’ve had this rash for three weeks and I have no idea where it came from. What is it?” You know? And, and you have no idea how to check that information or, or, you know, whether it’s correct or not. Uh, uh, that is a perfect, perfect example, and the electrician, uh, is even more perfect from my perspective because you can get electrocuted and you can die- Yeah.

if you’re playing around with electrical currents, right? It’s a, it’s a great analogy. Um, and you know, the rash situation, if you, if you’ve had poison ivy before and you’re using an over-the-counter medicine, right? It, is the generic okay? Um, you know, that’s very different than, I don’t know what this rash is.

I’m gonna [00:09:00] guess that it’s this. I’m gonna put something on it. Mm-hmm. And in fact, you could make it much, much worse. Yeah. Exactly. Um, so getting into the, the privacy perspective of things. You know, most consumers, we’ve gone to the doctor’s office. We’ve gotten the HIPAA forms, so we know about HIPAA, the, the privacy law that protects medical information.

Is medical information that is uploaded into just a regular AI like ChatGPT, is that protected by HIPAA? It’s not. And you know, it, it’s a, it’s a myth that every piece of health information is covered by HIPAA, and it’s a myth that any company that has health information has to follow HIPAA. That is a myth.

HIPAA is designed for very specific covered entities and business associates, and a generative AI tool doesn’t fall into, usually, into either of [00:10:00] those when it comes to the consumer inputting data into that tool. So when you’re inputting your most sensitive data, including your health information, your genetic information, um- That information you are giving freely and willingly to a private commercial enterprise that does not have to follow HIPAA.

And in fact, if you read the privacy policy of these gen AI tools, they will specifically tell you not to. Mm-hmm. They warn you against inputting health information, sensitive information, proprietary information. So, um, it, it is a misnomer and a myth that people have that laws will protect all of my health information.

Yeah. That is not true. Yeah, I mean, it’s the same thing as, you know, telling your friend about your healthcare [00:11:00] information. They’re not covered under HIPAA. Correct. So, you know, if HIPAA does not apply, um, could there be state laws that step in and prot- uh, provide protections? I mean, considering the fact that, um, you know, not every state has its own law.

Great point. Um, i- in the United States, we do not have a, an, a national, you know, privacy law, so lots of states are coming up with these, you know, different privacy laws. Uh, many of them include health information, but really only in the context of if there is a data breach, you have to notify the individual if there is a data breach and their health information was compromised.

Y- you, you know, a consumer inputting their health information freely, willingly is, is [00:12:00] not a data breach. It’s not an unauthorized disclosure of health information even though the company says, “Don’t do it.” So in a handful of, maybe double handful of, of states, if there was a, a compromise of that data, the company would have to notify, but that data is still there and is still being used to train.

Mm-hmm. And you consented to them collecting and, and using it. Yeah. So I, I know that was a long-winded answer, but, uh, the an- the, the true answer is I would not rely on state laws to protect your health information when you are freely and willingly uploading it into a third-party platform. Yeah. Yeah, absolutely.

Now, I know that there’s definitely some people doing this, um, who think they’re being clever. Um, you know, they remove their name, address, date of birth from some, like, test results from [00:13:00] the doctor’s office and then upload them into the AI. Is that enough to make that information anonymous?

No. So you need to remember that There, there, it’s very easy to re-identify information based on certain data elements. And so I don’t wanna talk about de-identification and anonymization, and there’s lots of different ways to slice and dice data. But what consumers need to understand is that just because you took your name, your address, and your date of birth out of a data set, these companies have so much data on all of us that it would be very easy to take other information that they can gather on the public web and otherwise to be able [00:14:00] to re-identify it.

It is a, it, it, at the very minimum, you should be deleting your name, your address, your date of birth. Don’t ever put your Social Security number in or any other, you know, uh, highly sensitive data. But the hospital name, the doctor’s name, all those things can… It’s very easy to re-identify. So- Yeah … um, that will not protect you, uh, from re-identification.

Yeah. I, I would also say, you know, you could be logging into your account on OpenAI and then inputting that information, and your account- T- … has your name, address, email. Carol. Um, you know, or- And your, and your IP address, right? Yeah. Yeah, and your device identifier and, and, and all of that information that’s just collected automatically, you know, they can very easily tell, um, who that is.

And even if you input into ChatGPT, like, “This is my friend’s medical record,” you know, you’re not really fooling anyone. Correct. [00:15:00] So, you know, what happens to this medical information when it’s uploaded into the AI? Um, what are the biggest privacy risks that we should be thinking about? Is it retention, training, sharing with third parties, selling, um, things like that?

All of them. I’m not sure I could prioritize all of those. But- … the, the biggest risk from my perspective is that, you know, your, when you input your sensitive data into a third-party platform that’s, that uses generative AI, they’re, they’re training off your data. So we talk about it going into a black box, and it really does go into a black box, and it’s used over, and over, and over, and over again.

And it can be used as output if someone puts a prompt in. So your actual identifiable data with your name, and your diagnosis, and your treatment, and all of the things that you, that you have queried [00:16:00] is now available to someone else as an output. So that is, from my perspective, one of the highest risks of Of inputting your health information into a Gen AI tool.

In addition, uh, I, I do think that, um, we know that the, uh, Gen AI platforms are selling the data, they’re disclosing the data. They have every right to do so. It is legal. You have consented to it. They can basically do whatever they want, provided that they are telling you what they’re doing in their privacy policy.

So before anyone inputs anything into a Gen AI platform, I highly recommend that you read the privacy policy to see what they are doing with your data. And most of them would say, “We’re doing whatever we want with our data, which is legal, so don’t put anything that you really care [00:17:00] about or that is sensitive or proprietary, because we’re gonna train, we’re gonna disclose it, we’re gonna slice and dice it, we’re gonna– we’re going to sell it.

We’re gonna do whatever we wanna do with it.” And so that’s why I think people don’t understand that- When you are inputting that data to- into a gen AI tool, they are using it any way they want, and they’re allowed to do that. That’s not a criticism of the, of the companies. It’s, it’s that they, um, they are being transparent and they’re telling you not to do certain things, and so that’s why it’s so important to read these privacy policies.

Yeah. So talking about training, um, these models. So let, let’s pose just a hypothetical. My name’s John Doe. I have Lyme disease. I tell the generative AI that I have Lyme disease, and that because of the Lyme disease, I’m feeling very tired every day. [00:18:00] Jane Smith goes into the same AI and says, “You know, I think I might have Lyme disease.

I’m feeling very tired every day. Do you have examples of other people who feel this way?” Could the generative AI say, “Oh yeah, John Doe- Yeah … you know, he feels very tired ’cause he has Lyme disease”? Absolutely. That’s a, that, that is, it absolutely happens because it is the same prompt, and the prompt is, “Who else is having these symptoms?”

So the, the algorithm is going and it, it not only is John Doe, but it could be hundreds of other people who had the exact same symptoms and the exact same prompt. Mm-hmm. And so now Jane Smith knows that you are tired and have Lyme disease. I don’t know about you, but I’m not sure I want Jane Smith to know that about me.

Right. Yeah. I, I think about how this could be used for nefarious purposes for, like, insurance [00:19:00] companies or employers or, you know, anything like that where those prompts can just basically become public knowledge, and you certainly didn’t intend for that to happen, but because you put that information into the AI, it did.

Yeah. There’s, the, the biggest risks of gen AI, as you well know, are, you know, not only the hallucinations and the privacy and the security concerns, but also bias discrimination- Mm-hmm … and the data that’s being used for, uh, other purposes. What I will say is, um, that there are laws that are popping up, and in fact, there’s guidance and best practices for the insurance industry, the healthcare industry, other regulated industries to, uh, put measures in place so that, uh, you know, they can’t use that data for a nefarious purpose.[00:20:00]

That said, it certainly could be used by threat actors in a nefarious way for blackmail, for, you know, for ransomware, for different- Mm-hmm … uh, ways that they can try to, uh, extort money from consumers. Yeah. Now let’s say I input this information into the AI and I just click delete on the conversation afterwards.

Does delete actually mean that it’s gone forever? So, um, i- in my understanding, uh, you can delete that prompt, but there, uh, it’s probably not completely deleted. There are very specific reasons why these platforms would need to keep the prompts. One would be for regulatory purposes, another could be for data retention purposes.

Um, so there’s probably details in the [00:21:00] privacy policy around how long they actually retain the, the prompt and the conversation. Um, but I would be suspect of that, again, not because it’s illegal and not because it’s, um, nefarious, but because there are business reasons why those prompts and those conversations need to be retained.

That said, um, I certainly w- would recommend that these prompts be retained only for the minimum amount necessary. Yeah, a lot of times when, um, the, the information becomes part of the training data set, it’s almost impossible to delete- It’s- … it as well. Um, which is really unfortunate because, you know, most privacy laws have the right to delete.

Um, so I’m not quite sure how these AI companies are actually gonna implement or [00:22:00] effectuate that right if it’s impossible to delete something from the training set. Yeah. I, I think, and I think it’s also pretty impossible to be able to track internally the training and use of that prompt and conversation.

Mm-hmm. Which, when it comes to the right to delete, is, I think gonna be very challenging. For sure. Um, so AI can, you know, make inferences, so maybe I give it a whole bunch of innocuous information. Um, and, you know, I might say, “Oh, I, I have a headache today. I’m gonna take an ibuprofen.” And then, you know, later on, “Oh, man, that’s so bright today,” you know, or, uh, “Sounds are kind of bothering me.”

Um, so it might infer that I have a particular medical condition, um, like let’s say I have migraines, um, from information that’s innocuous, like I didn’t tell it that I have migraines. Is that inference itself potentially sensitive [00:23:00] information? It, it could be, right? So if it’s making inferences about certain diagnoses, um, obviously the concern would be for a, a more, I don’t wanna say a more sensitive diagnosis, but, you know, someone has migraine might not, might not be that, uh, might not be that detrimental.

Um, but if you’re missing work because you, uh, you know, you have a headache and there’s light flashing behind your eyes, or if we look at other diagnoses that are, that are a lot more sensitive, you know? Oh, y- like if you’re pregnant, right? Mm-hmm. Or reproductive health, or family planning, or, you [00:24:00] know, other types of, um, you know, diseases that would be considered more sensitive.

I think, uh, sexually transmitted diseases, other kinds of things that, that are, uh, mental health, uh, psychological needs, I think that is, uh, is where I worry a lot about the inferences that are being made by, uh, gen AI, and the, and the correlation of those, those conversations to an actual diagnosis that could then be used as output.

So if they’re inferring that I’m pregnant but I’m not- That could be detrimental as an output. Mm-hmm. Yeah, and it, it’s very easy to get to those inferences, right? If I just tell Gen AI, “Oh man, I’m feeling kinda nauseous today. What kind of ginger ale do you recommend?” [00:25:00] And then a couple weeks later, “Oh, I just started this new birth control pill, but I missed one, and then how do I, what do I do now?”

And all those things can add up very, very quickly to an inference of, “Oh, you’re pregnant.” Yep. Yeah. It, it’s just crazy how, how fast we can get there. Yeah, and, and, you know, already before Gen AI, already companies are making- Mm-hmm … these inferences, right? The, the big social media platforms, the, the, you know, the platforms where we purchase items or the platforms where we’re watching movies, they make a lot of inferences about us.

Um, but they’re not, at least as far as I know, they’re not training algorithms and disclosing- No … that and training where it could be a, a prompt or an output elsewhere. Yeah. So this has been happening a long [00:26:00] time, right? You, you go and you, you stream a movie, and then all of a sudden the suggestions for you, or you go and purchase something from a, a well-known, uh, a well-known company, and they start saying, “Oh, but you know, you should probably consider purchasing this because of that.”

Mm-hmm. There have been in- inferences about us forever, but now the inference can be an output, and that’s what’s a little scarier. Yeah, absolutely. Um, so let’s do a privacy lightning round. Um, I’ll give you a couple examples, and, uh, you have to answer upload as in upload this into the AI, it’s fine; don’t upload; it depends, and why.

Um, so the first one is, “Here’s a picture of my rash. Uh, what do you think it is?” See, I, it’s not fair because I wouldn’t upload any of this. [00:27:00] Um, I think I, I, I wouldn’t upload it. Okay. It depends, I guess. Um, okay. I wouldn’t upload it. Okay. Uh, here’s my 600-page medical record, summarize it. I definitely wouldn’t upload that.

Okay. I’m 13 and I’m taking medication for my ADHD and I’m experiencing these symptoms. What could be causing my symptoms? So that, that is one that I, I also wouldn’t use AI for. I mean, that is just something that, uh, needs to go to a medical professional. Agreed. Um, my friend sent me their medical records and asked me to have AI summarize them.

A- again, I, I wouldn’t upload that, uh, for your friend. Um, I just think, um, one, I, I, I, I would probably try to counsel my friend about why no one should upload it, but I, I don’t think it would, um, [00:28:00] I don’t think it would be smart to upload your friend’s medical records. Agreed. And is it safe to take medication X with medication Y?

So this and the first one, right? Those are the it depends probably. Um- The, and again, I’m, I’m more conservative than others probably, but, you know, I, I do think that, um, I, I would, I would prefer that someone go to a, you know, a, a reputable medical, um, site, a website, or a, um, uh, a WebMD kind of, uh, site than into ChatGPT.

Um, I think to get general information about drug-to-drug interactions might be okay, but the problem with that is it might not be. And so I, I would counsel or I would recommend if that person go to [00:29:00] a, a medical website or call their physician. Their physician is the one who should be, um, advising the patient on drug-to-drug interactions.

Yeah. I would personally recommend talking to the pharmacist, because from my experience- Yeah … like most, uh, physicians don’t know interactions really well. Um, but pharmacists are, like- Yeah … absolute pros at that, so checking that, and I know that most pharmacies have like a very quick, like 30-minute appointment you can make for a rash or to ask questions and, you know, usually they’re pretty cheap and they, they can help out without…

Because taking medications that have unsafe interactions can like cause death. For, for sure. You know? Um, so you have to be careful. And so but again, it, it’s a medical professional, right? Yeah. It’s, it’s a professional not, not a, an algorithm. Yep. Yeah. For sure. Um, now, you know, as a consumer, I can just easily [00:30:00] choose not to put my information into the AI, right?

It’s something, it’s an action that I’m taking, I can just stop doing that. Are medical professionals using AI at all? Like, could medical professionals upload my protected health information into an AI? Uh, the answer is yes, and medical professionals are using AI. The difference is that medical professionals are using AI tools that are closed systems, right?

So they have a specific contract with the platform on what the platform can do with the data, how it can use the data, it can’t train. So they can use the benefit of summarizing vast amounts of data in a closed, safe, and secure environment. So absolutely, and we want medical professionals to use AI, but they’re not inputting your medical records into, [00:31:00] you know, ChatGPT.

They’re using very specific enterprise platforms where there are, um, privacy, security, contractual obligations, and safety considerations. Do they, like, as a patient, do they have to tell me that they’re using AI with my information or no? So, um, interesting question. Most, most medical professionals that I know are letting their patients know, especially if they’re using, like, a transcription service when you’re in the room with a medical professional.

Mm-hmm. Um, but in general, they don’t have to disclose to you that they’re using software tools because think about it, they’ve been using software tools for decades. Um, this is just one more, this is one more, uh, software tool that they’re using. What they have to tell you is that they are doing it [00:32:00] in a HIPAA-compliant manner, and they do that through their notice of privacy practices.

So you, you can feel comfort that if they are using generative AI tools, it will be part of their HIPAA compliance program, it’ll be vetted, and it’ll be compliant with HIPAA if they’re a covered entity. That makes sense. Now, you know, let’s say I go to my doctor’s office and they tell me, “Hey, I’m gonna be using this AI transcription tool.”

And, and let’s say as a patient I’m not comfortable with that, can I say no? Of course. You can of course say no. Um, it, it will make that physician’s day a little longer probably, and maybe interrupt the flow, but you can. You, you do not have to agree, and you do not have to, um, uh, y- you, you don’t have to consent for them to use that transcription service.

Good. Again, I would, I would- [00:33:00] I would caution people or give them comfort that that transcription tool or whatever AI tool they’re using, uh, should be, have been vetted, and should be HIPAA compliant. So it’s not like, it’s not like you dictating something into ChatGPT, which is not protected. This transcription service is part of, you know, should be HIPAA compliant and will stay within the, uh, the confines of the, of the hospital or the doctor’s office I, I’m curious, uh, if this situation has ever come up where, you know, some young intern has a lot of patients, and they take care of those patients, and they go home with all their notes, and they input that into ChatGPT.

Would that be a violation then, because they’re going outside of the, the approved hospital system? Yes, that would definitely be an unauthorized disclosure of [00:34:00] PHI if they were taking in- PH- protected health information from the hospital setting and putting it into a, some sort of a private platform. Yeah.

That would be a violation of the hospital’s policies, for sure, and probably a violation of HIPAA. Yeah, I’m gonna guess they would be immediately terminated if they did that. Well, a- as you probably know, all, um, all covered entities and business associates have to have a sanctions policy. So that would fall under the sanctions policy- Yeah

for sure. So y- final question for you. You know, has learning about AI and the privacy risks associated with it, has it changed how you use AI in your personal life, or, or whether you use it at all? So I, I do use AI. I use AI for very innocuous things, right? So I’m going on a trip, and I… My husband and I are going on a road trip, and I wanna know what are the best restaurants in all the different cities, [00:35:00] or what’s the best itinerary, or like, I don’t, you know, I, I’m fine with someone else knowing that I’m going on this road trip and where I might be going- Mm-hmm

to eat, right? Like, who cares? Yeah. Uh, I do not use it… A- and we, our law firm uses it, and we are, um, embracing the use of AI tools. Our clients are asking us to, and we’re doing it in a very cautious but, um, safe and secure way. And so I do use AI tools, uh, in my workplace as well. But again, it’s, it’s cl- a closed system, and it’s, it’s safe and secure.

Um, I am… Otherwise, I am very, very cautious about the information that I am giving to, uh, any gen AI tool. Um, much like I’m very cautious about giving any platform my information. I’m the one that [00:36:00] says, you know, “I decline all the cookies,” and I reject cookies. Mm-hmm. And that’s me, because I believe that our data is being, um, aggregated, and I, and, and, you know, we’re, we’re being targeted.

And I- P- some people are fine with that, I’m not. So I am probably more cautious, although on the things that I do use it for, the more innocuous, generalized things, it’s, it’s a great tool, right? Yeah. It’s amazing. It saves me a lot of time. Yeah. I’m the same way. Like, I’ll look it up… I’ll use it to look up, like, dinner recipes- Right

or what should I pack for lunch, um, you know, or i- it’s my sister-in-law’s birthday, can you write a greeting card for me, or something like that. Um, I do have to admit, I did use AI for medical use once. But I had to obtain my, um, medical records, and on the medical records, there was this phrase that devel- uh, that, like, was written over and over again [00:37:00] for every, uh, meeting, which was, “Patient is a well-developed female.”

And I’m like, “What the heck does this mean?” I was like, “What is this?” Well, you could take that, you could take that a couple different ways. Right. So I just, I didn’t upload the record. I just wrote in, like, you know, “This is what it says. What does it mean?” And apparently, it just means you’re, like, normal weight and your arms and legs are, like, the right length.

And I was like- Yeah … “Okay, that makes me feel so much better.” Right? But it is… It, it’s an old term that is used over and over and over. Right. And it’s like, I get you’re, like, you’re looking at it literally going, “Well, what does that mean?” Yeah. It could mean something else, right? Right. Yeah. I was getting really uncomfortable with seeing that.

But the AI explained to me that it’s just, like, part of the general template. Yeah. It just means, like, your arms, legs are the right length. Like- Yeah … you know what I mean? Um, but yeah, that’s- But that’s, that’s a good use of AI- Yeah … for, for a medical term, right? What does this mean? Yeah, exactly. So but yeah, I [00:38:00] definitely agree with you.

Being more cautious about this is important, um, especially since we don’t know exactly where this data goes, how it’s being used, who it’s being shared to, with, how… You know, who it’s being sold to, and how it could potentially be used against us. Yeah. I, I think caution is, is really… You know, safe and secure and, and baby steps is- Yeah

is really, from a consumer standpoint, I think the right way to go. We will… It, it’s just like social media platforms, right? We all used them at the beginning, and now- Mm-hmm … 15 years later, we understand what the, you know, detrimental effects it has had on youth and, and children, and I think we’re gonna, in 10 years, we’re gonna know a whole lot more about- Yeah

how these early days of the use of AI tools is going to affect all of us and society. And so I think, [00:39:00] um, especially for consumers, being educated, understanding, reading the privacy policies, and taking baby steps is really the right way to go. Yeah, absolutely. Linn, thanks so much for joining me on the podcast today.

It was my pleasure. It was a lot of fun. Thank you. Good. Um, and to our listeners, make sure to subscribe to Privacy Laws so that you don’t miss our next episode.