00:01We know that young people nationally are turning to chatbots for help, for care, for psychosocial
00:08care when they experience harmful interpersonal experiences. And then when they experience
00:14or if they experience an instance of online sexual victimisation, they then disclose that
00:20experience to an AI. We know that they're disclosing to AI at much higher rates than
00:26they are to authorities. And this is a whole new thing. We haven't seen this before.
00:29OK, so we know that what they're disclosing to these AI chatbots, but do we know what
00:34the AI chatbots are telling them? Because they must be going on there seeking advice.
00:39No, crucially, we don't know. We don't know what they're saying back. And that is a vital
00:44part of this conundrum. We know that they're seeking help. We know that they're going to
00:49AI for life advice even. We know that they're going there regularly for that kind of advice.
00:55And when they're victimised, they disclose to an AI. We don't know what's being said back.
01:00That is a crucial piece of safeguarding for young people.
01:03We know that given this latest research, one in three kids who are sexually victimised
01:08online are turning to AI chatbots. But more broadly, you touch on this, do we just know
01:12the magnitude of young people who are turning to AI chatbots for advice?
01:16Yeah, this is enormously important. It's part of a broader ecosystem of young people turning
01:22to AI. So we know that around one in three or a little over one in three seek personal
01:29advice from an AI. We know that more than one in five seek psychosocial counselling from
01:35an AI. And we know that more than one in 10 disclose hurtful experiences to an AI platform.
01:40And what's really important about that is that it's not just they're doing it, it's
01:45that they're doing it at least once a week. So nationally, young people are turning to
01:49AI platforms for advice, for help seeking and for talking about hurtful experiences, irrespective
01:56of whether they experience victimisation or not.
01:58What are some of the most obvious and immediate harms here that you see?
02:04The really crucial part that we see in our data is that when young people turn to an AI platform
02:10to disclose a harmful experience, less than half of them go on to tell another human. When
02:17someone approaches a human first, about half of them go on to tell another human. That secondary
02:24disclosure for help seeking is absolutely crucial. We don't know why young people, when they go
02:31to a chatbot first, aren't then going on to tell a human at such rates. But we really need
02:38to find that out because that is a real concern. And it's something that needs to be really
02:44carefully managed.
02:45What are some ideas around bridging that gap?
02:48So it's about pathways. It's about help seeking. We know that young people are going to AI chatbots
02:54at much higher rates than they're going to authorities like police counsellors, doctors. And that's
03:01largely because they might be finding out about their victimisation on their phone, on their
03:05computer. And they're flicking across to a chat GPT app to Claude and saying, this just happened
03:11to me. What should I do? Where can I get help? Ultimately, that's a clearer pathway. It's a quicker
03:17pathway than making an appointment with a doctor, getting a mental health plan, going and seeing a
03:21psychologist, which can take weeks, if not months. And so it's about pathways. It's about ensuring
03:27that that service access and that disclosure access is quick and easy. Crucially, it needs to be easy.
03:34Are there ever any benefits of seeking advice through a chat bot, through AI?
03:40Look, we know that the acute phase when someone's been victimised and they need help, we know
03:49how crucial that is and how important that is for a young person. Having the speed of access,
03:54being able to talk to and disclose to someone or an AI, that's a really important pathway.
04:01Obviously, maximising the rate at which people disclose to humans is incredibly important.
04:09But if it's well managed and if the responses from the chat bot are appropriate and they
04:14safeguard the young person, then it's a legitimate avenue for seeking help.
04:19We heard the AFP commissioners there saying it can't all be on police addressing this issue.
04:24How do you think family and community can step up here? What sort of tools might you be advocating
04:30for conversations they should be having?
04:33Look, I'm reluctant to displace responsibility further onto, for example, parents to be responsible
04:42for and have the in-depth understanding of management of really sensitive disclosures.
04:51Ultimately, those conversations are incredibly important. But any avenue that we can put, any
04:57support that we can put around young people is good support. Any disclosures are good disclosures
05:02if they ultimately end up in help. Community in terms of getting around kids, I think that
05:09there are programs that are really, really useful in supporting young people to disclose
05:16to the appropriate sources. And I think that that is vital until we know what chat bots are
05:22giving back to young people, until we know that those pathways and those supports are really
05:27appropriate and they really safeguard a young person. Until then, it is a responsibility
05:33of the people who are around the young person. But there is an opportunity here to give young
05:39people a quick, robust avenue to disclose and get help.
05:43Okay.
05:43.
05:44.
05:45.
