What can you do if you find an AI generated image of your child?

In one click, AI can generate a fake image of anyone. In one click, it can undress them. And AI does not care whether that person is a consenting adult or a child.

This episode is about one of the most urgent and underreported risks facing children right now: AI-generated deepfakes. Not as a future concern. As something happening today, in UK schools, to children the same age as yours.

Florence sits down with Tara Steele, founder of Safe AI for Children, a former intelligence officer with 13 years of experience in risk assessment, who now dedicates her work to shaping the AI world our children are growing up in. Tara's background in intelligence, law, and data protection makes her uniquely placed to answer the question every parent needs to ask : if this happens to my child, what do we actually do?

Together they cover what deepfakes actually are and the different levels of harm they cause, from children manipulating a friend's image as a joke, to AI-generated bullying, to the creation of child sexual abuse material from ordinary school photos. Tara explains clearly what is criminal, what is not, and what the practical first steps are in the first 24 hours if a parent discovers their child has been targeted.

They also address something that goes beyond the individual : why AI was rolled out without proper child safety testing, why the pharmaceutical and aviation industries would never have been allowed to operate this way, and why parents are being left on the front line of a problem they did not create and should not have to solve alone.

This is not an anti-AI episode. It is a conversation between two mothers who believe the momentum for change is here and that parents deserve to know exactly what they are facing and what they can do about it.

The show notes below include direct links to the Report Remove tool, the Take It Down tool, and Tara's resources at Safe AI for Children.

🎙️ Listen on Apple Podcasts : podcasts.apple.com/gb/podcast/the-ai-mum-podcast/id1892893748

 📱 Instagram — @theaimumpodcast
 🌐 Safe AI for Children — safeaiforchildren.org

SHOW NOTES

Guest: Tara Steele, Founder, Safe AI for Children. Former intelligence officer of 13 years. Background in AI governance, law, and data protection. Speaker at UK Parliament and UNESCO. Mother.

Website : safeaiforchildren.org

IF THIS HAS HAPPENED TO YOUR CHILD - ACT NOW

Step 1 - Support your child first
Make clear this is not their fault in any way. Do not restrict their devices as a response. Your support is the most important thing in the first moments.

Step 2 - Do not delete anything
Preserve evidence. Note URLs. Do not forward, share, or store images on other devices. Schools and safeguarding leads should follow the same protocol.

Step 3 - Contact law enforcement
If there is any immediate danger contact emergency services first. For non-emergency situations contact your local police. The creation, possession, and sharing of AI-generated sexual images of children is a criminal offence in the UK under the same legislation as non-AI images.

For reporting in the UK you can also contact the Internet Watch Foundation at iwf.org.uk, the organisation responsible for identifying and removing child sexual abuse material online.

Step 4 — Use the image removal tools

🔗 Report Remove - UK, run by Childline
For under-18s in the UK. Generates a digital fingerprint to prevent images being reuploaded across platforms.
childline.org.uk

🔗 Take It Down - worldwide
For anyone globally. Same digital fingerprinting approach to prevent images spreading online.
takeitdown.ncmec.org

Step 5 - Contact the platform
If the image is on a social media platform, report it directly. Platforms have a legal responsibility to remove it.

WHAT TARA WANTS EVERY PARENT TO KNOW ABOUT PHOTOS

Once a child's image is online, a parent and a child have lost control of it. A child cannot consent to their image being shared, and when they are older they may feel strongly that they should have been asked.

Tara's advice is that schools should urgently transition away from posting identifiable images of children online. 

For parents posting on their own social media, the risk is real. Realistic deepfakes of a specific child can be created using as few as 20 existing images in as little as 15 minutes.

🔗 Read Tara's full guidance on sharing children's photos online:
 safeaiforchildren.org/advice-for-schools-and-parents-on-sharing-childrens-photos-online

🔗 Read Tara's article on AI nudify apps and what parents need to know:
 safeaiforchildren.org/ai-nudify-apps-children

THE THREE LEVELS OF DEEPFAKE HARM EXPLAINED

Level 1 - Image manipulation without consent
A child's image is taken and altered, made to look silly, wearing different clothes, doing something embarrassing. Not a criminal offence in itself but not acceptable. Consent matters. Schools need to educate children that this is not okay.

Level 2 - Bullying and manipulation
AI-altered images used to humiliate, manipulate, or engage in low-level blackmail. Schools must intervene. Depending on context, elements may be reportable to police.

Level 3 - Child sexual abuse material
The creation of sexualised or nude images of children using AI is a criminal offence. Report to police immediately. Use the Report Remove and Take It Down tools. Contact the Internet Watch Foundation.

KEY QUOTES FROM THIS EPISODE

"Once their image is out there, you have lost control of it as a parent and as a child."  Tara Steele

"Tech isn't this almighty being where no one has a right to be protected from it. Most specifically when it comes to children, it's just unjustifiable, and it really needs to change."  Tara Steele

"What I am absolutely not saying is that this should be down to parents. It would never be expected for a doctor to prescribe a medication and then expect the parent to go home and research it fully and decide whether it's appropriate or dangerous for their child. We would never do that."  Tara Steele

"That narrative, that we just can't do anything about it, is only correct if we decide it's correct."  Tara Steele

RESOURCES FROM THIS EPISODE

🔗 Safe AI for Children - read more about AI safety for children and access Tara's guides
 safeaiforchildren.org

🔗 Report Remove - UK image removal tool for under-18s via Childline
 childline.org.uk

🔗 Take It Down - worldwide image removal tool
 takeitdown.ncmec.org

🔗 Internet Watch Foundation - report child sexual abuse material online
 iwf.org.uk

🔗 AI nudify apps - what parents need to know
 safeaiforchildren.org/ai-nudify-apps-children

🔗 Advice for schools and parents on sharing children's photos online
 safeaiforchildren.org/advice-for-schools-and-parents-on-sharing-childrens-photos-online

🔗 NSPCC - support for children and families
 nspcc.org.uk

🔗 Childline - free confidential support for children
 childline.org.uk or call 0800 1111

FINAL NOTE FROM FLORENCE

This episode is dedicated to every parent navigating this alone. You are not responsible for the situation we are in. But you are powerful enough to change it.

The momentum is here. Share this episode with one parent who needs it.

🎙️ Apple Podcasts — podcasts.apple.com/gb/podcast/the-ai-mum-podcast/id1892893748
 📱 Instagram — @theaimumpodcast
 🌐 Safe AI for Children — safeaiforchildren.org

Transcript

SPEAKER_01 0:00

Quite shocking that the approach with AI to date has really been roll it out, see what happens, deal with it. Tech isn't this almighty being where no one has a right to be protected from it, most specifically when it comes to children, it's just unjustifiable and it really needs to change.

SPEAKER_00 0:22

Just to make it clear, you or your organization are not anti-AI, neither am I.

SPEAKER_01 0:28

Once their image is out there, you have lost control of it as a parent and as a child. While I am saying parents need to know, here are the things parents can do. The reason I'm saying that is because of the situation that we are in right now. What I am absolutely not saying is that this should be down to parents. We could use like the pharmaceuticals example. It would never be expected for a doctor to prescribe a medication, for that medication to be available to children, and then to expect the parent to go home and research it fully and decide whether or not it's appropriate or dangerous for their child.

SPEAKER_00 1:13

Welcome to the AI Mom podcast. Today's episode is about one of the things AI is very quick and good at: generating fake images. In one click, you can upload your picture and make it look the way you want. Unfortunately, in one click, you can also get somebody naked. And AI doesn't care whether that is a consenting adult or a child. The easy access to defects is a serious risk our kids are exposed to. We all heard about the UK school that posted a picture of its female football team on their website and ended up with a sexualized video of them AI generated by a party that wanted to leverage it properly. So our question today is serious and urgent. How can we as parents protect our children from AI generative defects? We are joined by Tara Steele, founder of the Safe AI for Children, an organization created to shape the AI world in which our children will grow in. Tara's background in intelligence security, law and data protection makes her the best person to help us answer this question. Tara, thank you so much for being here with us today.

SPEAKER_01 2:25

Hi Florence, thank you so much for having me.

SPEAKER_00 2:28

I'm gonna start with my first question. I'm really interested in your background in intelligence, which is very unique, and how you got to found safe AI for children.

SPEAKER_01 2:42

Yes. Like you say, I used to be an intelligence officer, so that's like my main career background. That was for about 13 years or so. That was quite diverse work, as you can imagine, and had absolutely nothing to do with AI. And then I've realized over the years that it's actually very relevant in terms of the way I think and predict and see risks. And while that's very relevant to me as a former intelligence officer, I'd also say it's relevant to most people. So it's definitely not the case that it's only techie people who get to have a say in how AI is developed and goes forward and affects children and affects us all because that's the bottom line, it does affect us all, and it requires all different types of perspectives and views and backgrounds to have an input into that. So, anyway, moving on from being an intelligence officer, I later on became very interested in AI risks. I became interested in the AI risks to children specifically shortly after ChatGPT was released, because I was aware at that point that I hadn't foreseen some of the specific types of risks. So while I've been thinking about the risks AI could present to like us globally, it had been on quite a conceptual abstract level. And when ChatGPT came out, it really made me sit up and pay attention. So I did some AI governance studies. As part of that, we developed a project and I developed a kind of a blueprint for an organization that would focus on children and their safety. And then I basically went on to found that organization in real life. So that was the end of 2024. We've been going for a little while now. It's been really positively received. This is something that a lot of people are very concerned about now. And yeah, that's where we are today.

SPEAKER_00 4:47

That's amazing, and thanks for the work you're doing. We need it. Just to go back on the bridge between your intelligence job and what you're doing right now. How do you explain the fact that AI was launched without any testing?

SPEAKER_01 5:01

Clearly, for anyone who has sort of had any sort of background in risk assessment, it's kind of shocking that this was just put out there. When those risks are foreseeable, if you just take a short amount of time to think about what could happen. So, in terms of the you know, new defined tools, I obviously that can create a huge risk for children. In terms of the chatbots and AI companions, it might take a few moments longer to think about why that could be problematic for children. But there are very clear risks in terms of the emotional attachments that children can bond with these things, and then the dangerous advice that they can give. Quite shocking that the approach with AI to date has really been roll it out, see what happens, deal with it. Like that's a very, very irresponsible approach to all the risks, and that is the fundamental issue that needs to be changed going forward, really.

SPEAKER_00 6:04

Especially when it comes to children. You drew a parallel at some point in one of your articles that protecting children from AI is like protecting their immune system from some diseases we cannot control. That is a very uh regulated area. How did we end up in a world where children were not even considered?

SPEAKER_01 6:23

I mean, it's a shocking approach, and the results, the tragic results, are entirely predictable. And that is a really strange situation to end up in because that wouldn't happen with something like pharmaceuticals or aviation. We take responsible approaches there. So those approaches, those regulations, those standards and accountability, they massively reduce the risk of harms, and that's to children and to everybody else. And then there's clear accountability when harms occasionally do occur, despite those measures being in place, and then there are penalties for that. That's like a very clear framework that we have in place to protect everyone from harm. And yet we seem to always have had this weird mindset when it comes to tech and tech companies that it is somehow a different beast entirely. And there are, of course, differences and nuances within it, and approaches need to be tailored, but tech isn't like this almighty being where no one has a right to be protected from it, most specifically children. And it's just been really odd to witness that happening, whereby it's an industry that's treated so differently to all other industries, and then there's so much lobbying and financial power behind that to try and keep that model in place, which is very irresponsible of any companies that are doing that, particularly when it comes to children, it's just unjustifiable and it really needs to change.

SPEAKER_00 8:12

I mean, just to make it clear, you or your organization are not anti-AI, neither am I. And I use it. It's more, as we said, like particularly focusing on children who are vulnerable and are not ready for this change to make sure we accompany them in the best way.

SPEAKER_01 8:29

I suppose you could use the same analogies, like someone who was, you know, campaigning for greater safety in aviation or whatever. No one would think of saying, oh, you're anti-planes. And yet, sometimes, especially when it's very, very, very pro-AI people, they tend to regard this as an anti-AI approach. And it is certainly not. Like that's kind of ridiculous.

SPEAKER_00 8:54

Now you put it this way, it does sound even more ridiculous. Yeah. Um we dive in more on our topic. I just have one last background question about safe AI for children. Just if you could tell us what's the objective of the organization, how can it help parents generally? Yeah, of course.

SPEAKER_01 9:14

So the approach that we've taken so far was to focus on talking to educators and parents, most specifically educators. So we've been trying to reach out to them a lot, provide guides, provide information so that they are in a more empowered position. And then to a slightly lesser extent, directly to parents, like right now, want to talk to parents and let them know what they can do on a day-to-day basis. And then the other thing we've been doing as well is talking directly to policymakers like UK government, wider organizations, to get the actual policies in place to make things safer for children.

SPEAKER_00 9:57

So if we can maybe dive into our topic of the day, which are deep fakes. I mean, as a parent, two things worry me about AI and children. The potential to hack children's attachment security, which you mentioned before, and which is what led me to actually create this podcast and try to learn more about AI in general, but also the actual harm to children that can be created via AI. And deepfakes are the biggest example in my view. If a parent discovered tomorrow that someone had created an AI deepfake of their child, what can they legally do about it?

SPEAKER_01 10:32

Absolutely. So that if sort of implied that there'll be lots of several different levels to this kind of thing. So when we say deep fake, that can mean all sorts of different things. So it could mean images that that don't involve children, like um misinformation or sometimes referred to as fake news, and that in itself is a risk to children because it distorts their understanding of the world and their ability to make sense of things, all sorts of other things. But that is a type of deep fake that's a risk to children. But then we move on to images that are created of children. So, like you said, there can be a specific individual child having their image manipulated using AI. And one of the ways that can happen is, for example, other children doing something as a joke. So children may see it as quite harmless to take a friend's image and use AI to make them like be doing a silly dance or wearing a particular outfit or something. That's one of the things that can happen. And children need education to help them to be aware that that's not okay if they don't have that person's consent. I mean, you can certainly argue it's not okay anyway, for various reasons, but the big issue with that one is consent.

SPEAKER_00 11:55

I mean, today, even me, like I take a picture of my children all the time without asking them. I think it's so easy to take a picture. It's I think it's really a society issue just to start with pictures. And again, there's a world between adults taking pictures and children taking pictures of children or adults taking pictures of children.

SPEAKER_01 12:15

So even for that kind of low-level incident that I've talked about, where a child's image is taken and made to look different using AI without their consent, that shouldn't happen. It's not fair to do that, and it can be used not only as a joke, but also for bullying, even manipulation, you know, even like sort of if it's just a slightly silly one, like low-level kind of blackmail and manipulation, it's not a good thing, and children need that education. In terms of the consent issue when it comes to adults, we definitely need to remember that because not only are you actively, if you put an image out of your own child, and I understand, definitely understand the motivation to do that, like as a proud parent, especially when they've achieved something and you want to share it. But once their image is out there, you have lost control of it as a parent and as a child, even when it's a young child, I personally do think it's important to recognise that that child can't consent to their image being there, and when they get to like 15 years old or whatever, they might be very upset that as a parent you chose to do that. So it's a big issue of consent, and then the perhaps even bigger issue of just the risk that that presents to the child. So we talked about those sort of low-level teasing kind of things that children need education about, but of course, it goes further than that. So a child could also be made to look like they're doing something really embarrassing or like criminal, stealing something, and then it starts to be used in a bullying context, and that can have a massive impact on the child. And obviously, schools will need to intervene to manage that, it's very distressing for parents as well. It's just a horrible thing to happen to a child, and that can be done quite easily. So, in we'll talk more about that, but just in terms of their photo being out there, if their photo isn't available in the first place, or if it's only available in a very limited way, that risk has already been reduced. It's really that straightforward. It should not be for parents and children to have to do that. We should not be forced into that position. It's not anyone's fault that this technology is here, other than the people creating it and rolling it out. But that is the situation we're in. So not being there reduces the risk. I think both parents and educators and older children themselves need to be really aware of that.

SPEAKER_00 15:00

That applies for schools as well, like wherever a child picture is taken, not sharing it on the internet is the first step.

SPEAKER_01 15:08

Yeah, we would definitely advise schools to quite urgently transition away from posting children's images online. Um, definitely identifiable images, because that increases the risk of being able to identify children used for blackmail. But all images of children online now present a risk, and that's our strong advice to schools is to not put children's images online and where you can take them down. The one incident that I'm aware of in my own personal life that I have children and one of their friends was taken from a school photo that's available online. It was an old one and it wouldn't have been seen as a risk at the time it was put up, perhaps. There's no blame there, but that is where it came from. And schools need to be really, really aware of that now.

SPEAKER_00 15:59

Even if maybe it takes out some of the joys, the communication, the marketing, it's not worth putting our children at risk because of that. But then just to go back to the to the question about if it happens, what do we do? And maybe the personal example you touched on. Once it happens, can we do something about it?

SPEAKER_01 16:22

So, in terms of those, I suppose we could call them lower-level incidents, it would be a case of getting that taken down, not shared further, talking to the school, asking school to intervene, and supporting the person that that's been done to if they are feeling embarrassed by it or traumatized by it in any way, and most certainly that if they've experienced bullying, obviously that needs to be dealt with. In terms of the creation of child sexual abuse materials using AI, that's a criminal offence. And it's dealt with not only in the UK, but in many other jurisdictions as well. Under the law, it's dealt with in the same way as non-AI images of the same kind, which means that it's criminal to produce those, to share those, to keep those. In other jurisdictions, the same outcome applies, but it maybe is under slightly different legislation. But obviously, I can't speak to all the countries in the world, but by and large, they say it should be. So when that happens, the police do need to be involved because a criminal offence has been committed. In terms of what the response should be as a parent, first and foremost, if there is any immediate risk, like if this involves a blackmail situation or anything else you could think of where a child is in danger, then obviously emergency services would need to be contacted as well as all the other usual procedures. And that would be first part of call if there's any kind of immediate danger. First thing I would say is support the child, make sure the child knows that you're there, that you're gonna help them through this, that this is in no way their fault. That applies whether it's AI generated or not, even if they've in any way been involved in the creation, they need to know that this is not their fault and they're gonna be supported through it. Don't punish them in any way, like by saying you shouldn't have been on your phone anyway, anything like that, because this isn't the child's fault and they need support. Secondly, I was always saying contact contact law enforcement, even if it's not an emergency, that you then contact them via the non-emergency channels. If this is to do with other children, the child's school and the designated safeguarding leads would need to be involved. It would be important for them to know and they should know under their policies, not to replicate the image in any way, not to share it, not to store it on other devices, not to forward it on, not to look at it themselves, but also not to delete it to preserve the chain of evidence, and they can certainly need to note of URLs for the police, um, note any other people who are known to be involved, that kind of thing, so that all the information can be provided to the police. There are services available to make sure that the images don't travel further online and to get them taken down. So in the UK, there's the report remove tool, which is from via childline, which you can just go on to online, and then there's a similar one worldwide called the take it down tool. There's some similar, there's some legislation of that name in in America, the take it down out. This isn't the same thing, they're not related. So it's a take it down tool. So you can look to both of those to get the image taken down if if it's online, those first 24 hours. You could get the original source image taken down as well. Obviously, if it's on a social media platform or something like that, they have a responsibility to take it down to supporting the child, making sure it's reported to the relevant bodies, making sure the the people who are able to take it down are contacted, and most of all supporting the child involved.

SPEAKER_00 20:17

Yes, 100%. And maybe we can in the show notes put the relevant bodies, yes. Um contacts just to make it clear. So, do I understand that that is a clear area where it's more sexualized pictures of children, and that's a criminal offense, but anything before that, which is like more bullying or more just uh an not just but an AI generated picture of a child, this is not a criminal offense. You cannot do anything about it but talk to the school or talk to the person where it comes from, is that right?

SPEAKER_01 20:50

Yeah, I mean, obviously there could be things within the context of that situation that could be criminal or or borderline criminal that would need to be reported to the police. Certainly, if something like blackmail is involved, then that in itself is like a different legal area. But in terms of just creating a non-consensual image of a child, like say doing a body dance with without their consent, then then no. Children need a lot of education about the fact that that's not okay, they need support where it happens, where that's applicable. Schools need to be involved in terms of intervention, whenever there's a bullying involved, anything like that, and that needs to be taken in its own, you know, there's the processes that are related to that. But no, in terms of criminal law, no.

SPEAKER_00 21:41

Good to know, just to understand where is the limit, and basically where also parents have more to do and schools to educate than take action, like as the criminal path you were describing. I just have one question. I mean, we live in this very intelligent world right now where. You know, you can do so many things with AI, like this type of pictures. As a parent, we have all these parental controls that are not always very effective. Is there a way, is there an app or something where you can know that your child's picture is being used in the wrong way?

SPEAKER_01 22:20

Um, I'm not aware of anything in terms of matching up like monitoring for an image of a specific child. However, there are measures coming into place legally to help prevent this from happening. So, for example, in the UK, there'll be legislation coming in to stop children being able to share nude images at a device level. Oh, that's interesting. Oh, and also I should mention AI itself being used to kind of monitor for child sexual abuse materials on the internet and get that taken down and hold people accountable for that. But in terms of matching up a specific child with a monitoring process, definitely nothing I'm aware of.

SPEAKER_00 23:12

Yes. I just think that the AI platforms should be in the same way, like I heard last time that Claude is designed to know if it's a child using it or they are above 18, even though you have you have to show your age in the beginning, they still look at it. So I think there should be something embedded within the AI chatbots where you're putting a picture of a child, it should stop you. Recognizing the the image as a child, yeah, as opposed to a exactly, and just not accepting to do any change on it. Does that exist or not yet?

SPEAKER_01 23:52

So it definitely exists in some contexts, like there are tools in other areas where you would have to do like facial recognition initially, and it can detect whether it's a child or or an adult. Legally, like there are concerns around privacy about that kind of thing, but there are also um good technical interventions to manage those issues. So, yes, potentially, definitely possible to stop this happening with children if the will is there and if technology is come, if that's uh enforced on technology companies. I would also say the one challenge is that even when this is enforced on bigger providers, big technology companies, that's still not gonna take away the risk because obviously now there will have been tools developed by people who are intentionally using it in a criminal way. They're not just gonna therefore put those kind of measures on their apps.

SPEAKER_00 24:53

I still do think that like the big platforms should lead by example. You see what I mean? And also the regulators would promote these things and not be defeatist because my personal view is that AI can be designed to be better used. Yes, and that's really critical at the moment, and I think that's what you're trying to do, that's what everyone is trying to do, is to shape the direction AI is taking.

SPEAKER_01 25:21

And I think most parents will feel that way and will support sensible measures to protect children. So the issue, one of the issues rather, at the moment is education so that parents actually know of the risk and political will for change so that governments act quickly to make sure that those measures are in place. And that picture is changing, but the sort of societal political momentum isn't quite there yet, and that's where we need to get to quite urgently.

SPEAKER_00 25:55

And why isn't it there yet? I mean, we had the social media wave, and now we have bands of phones at school. I mean, it is something very real, and we already had some unfortunate and sad issues with AI, with like teenagers committing suicide because of the hallucination that comes with it. It is very real, it is very present. Um why aren't authorities doing more? You talk to the parliament, to the UNESCO, why don't we have more mobilization and real action? Yeah, absolutely.

SPEAKER_01 26:33

I mean, there are several reasons for that. Like I spoke about the political will for change, like at a government level, that's changing. It's still not enough. But if I look back a few years ago and I was talking to, say, a politician and would go through these risks, they would generally say, you know, not in these words, but the gist of what they would say was, yes, you're right, but there's there's no will for that, it's not what people are asking for, it's not going to go to the top of the priority list because that's how this system works. Now it's not good enough, but that was generally the answer I was getting. It's different now. Politicians, policymakers are aware of the risks, are recognizing the right thing to do. And one of the issues that they face is pushback from big tech and lobbying and the huge amounts of money involved in that. So that's one aspect of it. Then again, in terms of parental awareness, although you are completely right in what you say that harms have already now taken place. This isn't saying this might happen, this might happen. It's now saying this has happened and this is going to happen more. Even though that's the case, this is still quite a new issue, relatively speaking, in comparison to social media. So we're only talking about like, I don't know, two, two, full years of LLMs like Chat GPT, image generation tours, a little bit later than that. And although to someone like me working in this area, that seems like it's been they've been around for ages, there have been so many harms done. We need to move so much more quickly, which is the case, which is true. To people who have like busy lives in other professions and looking after kids, you know, like life's a blur. I know it it just still might not be on their radar, so so the momentum still isn't quite there, and the education is so desperately needed. I was going to say about the social media issues, I think they could be helpful in terms of like riding that wave of momentum because the momentum is now there to create change in that area. What I hope will happen, and what I would like to try and influence happening is that AI measures are put in place alongside that. Um, that's not clear yet that it will be, but I do think it's entirely feasible and possible. What we need to make sure doesn't happen is that social media is no longer available to younger children, which in itself I'm definitely not saying is a bad thing, but what I want to make sure doesn't happen is that therefore they pick an AR AI for that to happen because AI companions chose a huge, huge risk to children. So it would be terrible if that risk got replaced with a greater risk of AI to children.

SPEAKER_00 29:38

I think it's so important, and that's what this podcast is about, and what you're doing is to give parents a voice. In order to give them a voice, they need to understand what we're talking about and what AI is. And as you said, not everyone knows. I mean, I heard a story like last time of a mom who lost her child that was more in 2024 to uh character.ai uh and she just didn't know what it was, like she didn't know at me too, like two years ago. I didn't know like this is this all came up to me while I was on mat leave reading more of the news, you know. So I I didn't know this was coming. That's and empowering parents to say that you can do something about it. Now is the time to shape things around AI for the children. We don't need to wait for 10 years like we did on social media. And it's not about banning it, it's about using it together, using it the right way, protecting the kids, having ideas as well. Like another topic I think about is having a mark that shows every AI generated picture is AI generated, like an obligation. Those are things the government can do.

SPEAKER_01 30:53

For example, the case that you're speaking about there, where the mum didn't know. That is the position I would have been in in that situation if I didn't work in this area. And it's incredibly important that no blame is attributed to those parents because they're entirely blameless and they just didn't know, and they were trying just trying their best to support their children. And there's truly, truly horrific incidents that have happened. And I would extend that further to say while I am saying parents need to know, here are the things parents can do, you know, give their education to parents so they can protect their children. Like the reason I'm saying that is because of the situation that we are in right now. What I am absolutely not saying is that this should be down to parents.

SPEAKER_00 31:47

Cyberpress is so important.

SPEAKER_01 31:49

Like again, we could use like the pharmaceuticals example, it would never be expected for a doctor to prescribe a medication, for that medication to be available to children, and then to expect the parent to go home and research it fully and decide whether or not it's appropriate or dangerous for their child. We would never do that. So while I'm sitting here using that same kind of approach for AIRIS, the reason I'm doing that is because of the situation that we're in and recognising that reality. It's not to say this is all down to parents. Parents need to take responsibility. You know, that that's a terrible narrative.

SPEAKER_00 32:28

And thank you so much for that. And it does make me feel better already. But it's not because you're right, it's more, as you said, it's more the situation we live with and the urgency, it's because no one else is doing it. I mean, my objective with this podcast was to learn myself before my kids are old enough to use AI. You see what I mean? Because I felt the urgency to do it, and as always, parents are left without any guideline, you know, about how to navigate these things.

SPEAKER_01 32:57

So yeah, and forced to the front line. Yes, and it's not fair to be in that situation, but it's not fair to children or to parents, but here we are.

SPEAKER_00 33:08

So but at the same time, I have to say that everything works with incentives, and we are incentivized to do something about it much more than any government or company, and that's the real situation we live in. That incentives are not the right ones at the company level and the government level. Um that's why unfortunately it we are responsible to have a voice. That's what I think personally.

SPEAKER_01 33:42

Yes, yeah, I totally understand. And the incentives for companies rolling out AI tools are a huge part of the problem. So, for example, with AI chatbots and AI companions, the companies are building those for increased engagement with the user because of the business model. And the ways that a lot of them are doing that is by actively putting things in place where the user is more likely to form an emotional bond with the AI model, and that can be hugely harmful, it's a huge risk, and the reason it's come about is because of the incentives when those models are um rolled out and developed. So, yeah, incentives are, I guess, what this boils down to and the challenges ahead to change those.

SPEAKER_00 34:41

Yes, I mean 100%. And that's why we need to know about it to be able to maybe influence with AI child, but our kids use uh what do they do with their pictures or things like that. So that's that's kind of the objective.

SPEAKER_01 34:56

It's a really important objective, it's just essential in the moment we're in. Parents have a right to know what's happening and what the risks are to their children and how they can help to reduce those risks where possible.

SPEAKER_00 35:11

Yes, 100%. I mean, this takes us to my last question that I asked all my guests. And probably we spoke about it through the episode, but we are not anti-AI here. We are just very aware of the harms that some of it could create on our children. I know what is maybe, in your view, the biggest downside of AI for children, but I will let you go on with your view and what is the biggest reason for hope at the moment.

SPEAKER_01 35:39

So, in terms of the biggest downside for children, it would actually be difficult for me to pick one and really rank the harms because some of them are so significant in their own way, it would be impossible for me to say that one is more of a risk than the other. So, for example, we're talking about creation of child sexual abuse materials, which is horrific, and we're also talking about chatbots which are built in such a way that they can cause emotional attachment, which has in multiple cases now led to children taking their own lives. So I can't like say I can't pick one thing, but those are two of the key risks to children, and there are more risks than that that we could go on to, like uh impact on their learning, their cognitive development, but those are the two we're focused on. And I would sort of encourage parents, if they can, to read a little bit more on those risks in terms of chatbots and emotional attachment. We have some materials on our website, safeaiforchildren.org, and I will be putting more on um and especially more on about the nudification tools and things like that. So it's a resource you can go to. In terms of what gives me hope. So, as we've already alluded to, there is some hope around AI for good, in terms of AI detection tools, for example, and get in in reducing the risk of these images, so AI can be used in that way for good. Also, like in a general sense where AI tools are very specifically built and targeted in like healthcare to improve outcomes, detect diseases earlier. Like, who wouldn't want that? So, those are positive sort of use cases for AI, and then in terms of hope in general, there can be a real sense of inevitability around this and around that we just can't do anything about it. We're just gonna have to manage these harms as they come. And my hope is that more and more people are understanding that that's an incorrect narrative. That narrative is only correct if we decide it's correct, and we do have power collectively to turn that around because there are so many of us that see how damaging these things can be that can create change, and we have to hold on to that hope that it will.

SPEAKER_00 38:16

Thank you so much. I do agree with you. The momentum is here, so we need to stick to it and do our best and be aware of the risk you mentioned. And I will share also in the show notes the links to the articles you mentioned. Thank you so much, Tara, for your time and this very hard-to-hold conversation from two months. It was helpful for me, for sure.

SPEAKER_01 38:38

Thank you so much for what you're doing. It's really valuable and for having me.

SPEAKER_00 38:46

Thank you for listening to this episode. I hope you liked it as much as I did. And it was helpful for you before we start the summer. As kids are on holidays, I am going to see you again in September with new ideas and new topics to discuss. Please share if there's anything you'd like me to focus on. And in the meantime, have a lovely summer.