Police are pleading with tech companies to let them use AI chat bots on their platforms to hunt for terror and criminal networks trying to recruit Aussie kids.
The move comes amid revelations children are hurting their siblings, killing their pets and being hired by criminal groups online, in a stepping stone to radicalisation and even terrorism.
AFP Commissioner Krissy Barrett told The Sunday Mail law enforcement was struggling to keep up with the pace of criminals preying on young and vulnerable people online.
“It really feels like our kids are at a crossroads,” Commissioner Barrett said.
Groups are paying middle men to hire children as part of “criminal gig economy” but also targeting them through online networks and forcing kids into sadistic online exploitation or violence through blackmail and manipulation.
Artificial intelligence is being used to find victims and “echo chamber” algorithms are taking children down rabbit holes to extremism and violence.
The chat bots could potentially flag extremist conversations such as bomb-making plans, terror attacks and school shootings, alerting authorities.
Commissioner Barrett has personally spoken with one tech company with further discussions with others planned at the Five Eyes Law Enforcement Group meeting this week to discuss working with law enforcement to combat criminal syndicates and state actors.
“We are seeing growing numbers of possession of violent extremist material and we know the impact that can have on a vulnerable young person in terms of desensitising violence and extremism,” Commissioner Barrett said.
Children in the vulnerable age brack from 13 to 16 are being hunted by online terror recruiters.
“And indeed we’ve then seen some young people mobilise towards acts of violent extremism or even terrorism.
“Again, the sadistic online exploitation, we are seeing again that same age bracket – so you know 13, 14 to 16 years of age when kids really are at their most vulnerable around trying to form their views and who they are in the world.”
The AFP has investigated 63 children aged under 17, with 33 of them charged with state or Commonwealth offences since 2020.
The youngest of these children were 12.
Ten cases were in Queensland, 32 in NSW, seven in the ACT, five in Victoria, four in South Australia, four in Western Australia and one in the Northern Territory.
Queensland’s Counter Terrorism Investigative Group arrested a 13-year-old Maryborough boy in May they allege was planning to kill children at a local school.
Detectives said they believed there was an “imminent threat” of an attack with the boy “well advanced in his intention to cause death”.
He was allegedly inspired by a Russian school stabber and school shooters, and watched videos prompting him to buy identical items on an online marketplace worn by those offenders.
In a separate case, information sharing between AFP and Five Eyes partners led to authorities stopping an alleged planned terrorist act in Australia, with a young person advocating nationalist and racist violent ideology on online platforms.
Investigators say the child allegedly planned to harm a stranger because of their racial background. His arrest led to investigators finding two other young people in Australia allegedly accessing violent extremist content and international partners also disrupted other young people overseas.
In another case, a young person allegedly discussed plans on an online platform about a potential shooting and terror attack. Joint Counter Terrorism Team investigators allegedly found material linked to an online sextortion network and his devices showed the youth had authored two manifestos and created an attack plan.
Investigators have seen a shift from “caliphate era” to targeted radicalisation through algorithm “echo chambers” of propaganda, as well as groups targeting youths through gaming platforms.
About 46 per cent of the AFP children cases investigated had an ideology breakdown of religiously-motivated violent extremism, while 24 per cent had ideologically motivated beliefs.
Investigators have charged 22 youths with violent extremism material offences since offences came into effect in January 2024.
The Australian Federal Police set up Taskforce Popilid last year to target online cybercriminal networks – mostly made up of young men – involved in crypto theft, sextortion, sadistic online exploitation and “violence as a service” which includes people being blackmailed into committing crimes.
AFP collaboration with law enforcement partners has led to 17 arrests, including five in Australia and 12 overseas.
“These online crime networks are glorifying sadistic online exploitation, cyber-attacks and violence,” Commissioner Barrett said.
“Right now, kids are hurting their siblings, killing their pets or injuring themselves because they are being blackmailed by other youth, teens or adults who enjoy watching the pain of others or are dominating and exploiting the vulnerable.”
Statistics also show that the AFP-led Australian Centre To Counter Child Exploitation received 82,764 reports of online child exploitation in 2024-25, with 76,708 already recorded this financial year.
There have also been 1675 sextortion reports received in the financial year to March.
In cases of sadistic online exploitation the AFP says cases include an Australian-based offender who coerced a child overseas to physically hurt a younger sibling. Authorities ensure the children were removed from harm.
The AFP is also investigating cases of Australian children hurting themselves, including by cutting themselves, after they were preyed on by members of a decentralised online crime network. The AFP says investigations are ongoing.
The AFP is also investigating cases involving possession of images and video depicting gruesome harm to animals, although at this stage these cases do not involve actual harm to animals by the Australian suspects.
There are fears that children being radicalised online, or paid to commit crimes by middlemen they met online, could lead to further serious and devastating crimes.
The Australian Security Intelligence Organisation has linked the Iranian regime to two antisemitic attacks in 2024, with one on Melbourne’s Adass Israel synagogue and another attack on a Jewish bakery in Sydney.
Children in other countries have been targeted by state actors in the United Kingdom and in Europe to commit offences.
Commissioner Barrett said it would be naive for law enforcement to not think children in Australia could be approached by state actors or proxies.
The AFP Commissioner said she and the Australian Criminal Intelligence Commission CEO Heather Cook aim to spearhead new tech partnerships and AI solutions at the Five Eyes Law Enforcement Group meeting in London this week.
Commissioner Barrett said law enforcement needed to work with tech companies, particularly using AI in a more “offensive way”, including using AI chatbots to deflect and overwhelm groups.
Chat bots are already used in the war against scammers, with bots intercepting and trolling cyber fraudsters and engaging in lengthy conversations, while also providing information about the groups to their associated sectors such as banking, and law enforcement.
“The other thing we’re particularly keen to explore is how do we use AI to pick up sentiment when it gets to a level that we feel like it’s really heading down the path of extremism or exploitation or the like, and divert individuals away from that,” Commissioner Barrett said.
“My concern is that without some serious intervention here, there is a real risk of particular harm to our kids but at a greater level social and economic fracturing in our community.
“My strong concern at the moment is traditional policing approaches can’t keep pace.
“But you know, I will be pushing that the problem is well and truly upon us. We need to move as fast as we can.
“I obviously acknowledge the privacy aspects of AI and that’s why I say we have very robust governance around how we use AI,” she said.
Commissioner Barrett said there would always be a human element or overlap with any AI technology used.
“But really in practicality, if I use the example of a chat group where we see conversations going down a path of extremism or starting to talk about preparing for an act of violence or terrorism – ‘I’m going to build a bomb or I’m going to build a gun and go and do a school shooting’ – these AI bots can pick that up the sentiment up in those conversations,” she said.
“And then can either deflect that straight to a human being who can intervene or as I said depending on the scale refer someone to educational material or referral pathways.”
Another example of a chatbot being used would be against criminal organisations using AI bots looking for people to take up a criminal contract or task, in a form of “criminal air tasking”, where crime groups decentralise their risk by outsourcing work
“We could just use AI bots to overwhelm that,” Commissioner Barrett said.
“So they don’t know who is a real human and who is not.”
Commissioner Barrett said children had a genuine curiosity to form views and in
instances felt they couldn’t ask teachers or parents, possibly because they were scared they were asking controversial or sensitive questions.
“So more and more are seeking answers online and you combine that with what we know is happening online at that rapid pace, in terms of you very quickly type an innocent question and we see literally within hours you can be down a deep dark rabbit hole.”
This article contains features which are only available in the web versionTake me there