Australian Police Develop AI Tool to Decode Criminal Emoji Slang and Combat Online Predators

Australia's police to use AI to decode criminals' emoji slang to curb online crime -- "crimefluencers" will be decoded and translated for investigators

Artificial intelligence has taken the world by storm, incorporating itself knee-deep in various aspects of our society seemingly overnight, for better or for worse. In the case of Australia's police, the intention is pure, and the goal is to prevent crime or catch suspects more effectively with the use of an AI that can translate modern, online slang, in particular, emojis. Young adults who are part of hate groups on social media that spread violent rhetoric and often act on it -- who Australian police refer to as "crimefluencers" -- are the main target of this new strategy. According to The Register, Australia Federal Police (AFP) commissioner Krissy Barrett spoke to the press today, highlighting how many Gen Z and Gen Alpha individuals are lured into these spaces, pushed to perform an act of tribute just to get in, such as recording self-harm. Once part of these decentralized communities, the crimefluencers coordinate real-world attacks using emojis that are difficult to interpret for investigators. "They are 'crimefluencers', and are motivated by anarchy and hurting others, with most of their victims pre-teen or teenage girls," said AFP commissioner Krissy Barrett. So, how does artificial intelligence help here? While we don't have exact details, AFP says it's developing a "prototype AI tool that will interpret emojis and Gen Z and Alpha slang in encrypted communications and chat groups," according to commissioner Barrett. It would likely have to be a multimodal natural language model (NLM) designed to interpret semantics as distinct, individual units, then use context awareness to determine whether something is an actual clue or just harmless slang. For instance, the skull emoji traditionally represents death or murder, but these days on the internet, it's gained more notoriety as "dying of laughter." The pizza emoji could literally mean pizza -- or, in some circles, act as code for a drug drop. This is where language embeddings come into play. If the model sees "(pizza) drop tonight?" paired with a blood emoji or certain slang terms, it could separate ordinary messages from ones actually worth flagging. To achieve this, the AI would need to be trained on a lot of open-source social media data from platforms like Instagram and TikTok, along with internal datasets from previous investigations and synthetic chats constructed to mimic online behavior. With transformer-based NLP models like BERT, it could gradually learn how meanings shift over time and, in turn, recognize when an emoji or slang phrase actually signals something dangerous. Since these groups have no unified leadership but share common interests in violence, nihilism, and sadism, with some of them even being Neo-Nazis, simply tracking down the people in charge is not going to reveal the chain beneath. The community has no center, so it has to be monitored on an individual basis, treating every possible threat with the utmost caution, and quickly working to prevent it. The AI is just one part of that; several other efforts are ongoing with different teams exploring ways to more effectively combat new-age crime. AFP has already joined forces with the "Five Eyes Law Enforcement Group" -- consisting of agencies from the US, UK, Canada, and New Zealand -- and has already arrested multiple suspects aged between 17 and 20, out of the 59 total that were flagged. Barrett also shared several cases where good ol' forensics still outsmarted computers, shining light on a recent crypto-laundering bust where a data scientist manually cracked an encoded recovery phrase worth millions after they spotted a pattern AI had missed. "While computer power is essential," she said, "it's not always as creative and innovative as a human." AFP is also experimenting with unorthodox forensics techniques, like trying to use cell phones buried with dead bodies to determine the age, cause, and method of death. The main goal is to estimate how long the body has been underground, and this is a collaborative effort with the University of Technology Sydney. Introducing the AI tool is simply the next phase of Australia's vision to control crime on a larger scale, folding in more tech and science as crime naturally continues to evolve.

Australian police build AI to translate emoji used by crims

Five Eyes intel alliance has created a team to target these scum who prey on kids Australia's Federal Police (AFP) is working on an AI to interpret emojis and the slang used online by Generation Z and Generation Alpha, so it can understand them when they discuss crime online. AFP commissioner Krissy Barrett revealed the plan in a Wednesday in which she discussed "decentralised online crime networks and loosely affiliated individuals in Australia and offshore who are glorifying crime online, such as sadistic online exploitation, cyber-attacks and violence." "While these networks do not have a centralised hierarchy or a single ideology, they are prolific and are attracted to violent extremism, nihilism, sadism, Nazism and satanism," she said. "They are 'crimefluencers', and are motivated by anarchy and hurting others, with most of their victims pre-teen or teenage girls." The groups are a matter for Australia's feds, she said, because "If it used to take a village to raise a child, because of advances in technology, it now takes a country to keep them safe." Barrett said these groups recruit members who "often have to pass a test or undertake a task, such as providing videos of the self-harm of others, or other gory content." "In this new, twisted type of gamification, perpetrators reach a status or new level in their group when they provide more content showing more extreme acts of depravity and sadism." Investigators identified 59 alleged offenders who are thought to be members of some of these prolific crime networks, and secured "nine international and three domestic arrests." Those arrested in Australia were aged between 17 and 20. "A sub-group within the Five Eyes Law Enforcement Group has also been established to target these groups," Barrett said. That means law enforcement across Australia, the UK, USA, Canada, and New Zealand are trying to stop these criminal scum. The AFP has set up a task force to target the gangs, and Barrett said it's "developing a prototype AI tool that will interpret emojis and Gen Z and Alpha slang in encrypted communications and chat groups to identify sadistic online exploitation." Which sounds like one use of AI nobody can object to. Barrett also discussed the AFP's anti-terrorism activities, which have since 2020 seen 48 youths aged between 12 and 17 investigated, and 25 charged with one or more terrorism-related offence. The commissioner mentioned one case in which a 14-year-old who had access to firearms and explosives, and a "nationalist and racist violent extremist ideology" used "a Snapchat account to post issue-motivated violent extremism." A tipoff to the AFP alleged the child planned a school shooting, and investigators thwarted that plot and secured a conviction. The Commissioner said posts of that sort are not unusual. "In most cases, social media or online gaming have been used to radicalise, or for a young person to flag their intent to commit ... a terrorist act." Another anecdote Barrett shared concerned an investigation into "a well-connected alleged criminal, who we suspected had stockpiled cryptocurrency from selling a tech-type product to alleged criminals around the world." AFP officers obtained a warrant to search the suspect's home, and found "an image on his device displaying random numbers and words. The numbers were divided into groups of six and there were more than 50 variations of the number groups." "Our digital forensics team determined it could be related to a crypto wallet - but the accused allegedly refused to hand over his passwords for his crypto wallet, which is a Commonwealth offence that carries a penalty of 10 years' imprisonment," Barrett said. "We knew if we couldn't open the crypto wallet, and if the alleged offender was sentenced, upon release he would leave prison a multi-millionaire - all from the profits of organised crime." While computer power is essential, is not always as creative and innovative as a human An AFP data scientist investigated the numbers, guessed that "some of the number strings felt wrong and they looked like they were not computer generated" and then decided to remove the first number in each sequence. That approach led to decipherment of a 24-word recovery seed phrase and recovery of $9 million in cryptocurrency. "And it was not a one-off for this member," Barrett said. "He recently did it again with another wallet, using different methodology, recovering more than $3 million in crypto." "To me this highlights that while computer power is essential, is not always as creative and innovative as a human," she said. Barrett's last anecdote concerned the AFP's forensics team which together with Sydney's University of Technology and "a pioneering body donation centre to understand how smart devices, like phones and watches, can help determine how long a body, dead or alive, has been buried during a natural disaster or foul play." "It is not as simple as it sounds because our scientists need to determine how soils, fluids and other contaminants can affect smart devices and the data they provide, such as heart beats or body temperature changes," the commissioner said. Barrett mentioned that effort in the context of the speech's theme that the AFP has developed skills and capabilities to tackle crime wherever it happens, but the speech also outlined real-world action including deployments to the Colombian jungle to destroy cocaine production facilities. ®

Australian police design AI tool to decipher predators' Gen Z slang

Australian police are collaborating with Microsoft to develop an AI prototype capable of deciphering Gen Z slang and emojis used by online predators. This tool aims to accelerate the identification and rescue of children from exploitation and radicalization. The initiative comes as Australia prepares to mandate social media platforms remove users under 16. Australian police are working on an AI prototype that will help them decipher Gen Z slang and emoji-laden messages written by online predators, a top official said Wednesday. Australian Federal Police Commissioner Krissy Barrett said social media had become a breeding ground for bullying, sexual exploitation and radicalisation. Police were working with software giant Microsoft to develop a tool that would unravel sinister messages hidden by seemingly innocuous emojis and slang, she said. "Clever AFP members, with Microsoft, are developing a prototype AI tool that will interpret emojis and Gen Z-and-Alpha slang in encrypted communications. "This prototype aims to make it quicker for our teams to save children from harm much earlier." Barrett also warned about the rise of so-called "crimefluencers" -- online predators who used their social media savvy to target young and vulnerable users. "They are crimefluencers, and they are motivated by anarchy and hurting others, with most of their victims pre-teen or teenage girls," she said. Australia will from December 10 force social media platforms such as Facebook, Instagram and TikTok to remove users under the age of 16. There is keen interest in whether Australia's sweeping restrictions can work, as regulators around the globe wrestle with the dangers of social media.