A man in Sydney, Australia was arrested this week after telling Anthropic’s Claude that he wanted to kill children at a daycare near his home.
As the Sydney Morning Herald reported, 29-year-old David Lan Thackray was taken into custody on Sunday after his disturbing interactions with Anthropic’s Claude were flagged by Interpol. He was charged with using a carriage service to harass, menace, or offend, an Australian criminal law.
Per the Herald, prosecutors cited documents showing that Thackray told Claude that he wanted to “go down to the local Westfield right now and shoot up the local daycare.” It’s unclear how Interpol gained access to Thackray’s chats and whether Anthropic sent them to the international policing organization.
In addition to openly discussing a massacre on a local daycare facility, Thackray’s conversations reportedly included discussion of Naveed Akram, the alleged terrorist accused of killing 15 people at a Hanukkah celebration at Sydney’s Bondi Beach in December 2025.
“What he intends to do is start murdering people in Sydney just like the [alleged] terrorist did in Bondi,” prosecutors argued in court, per the Herald. “We’re understanding what’s in his mind with his interactions with the chatbot.”
Thackray’s lawyer, meanwhile, reportedly argued that the 29-year-old was simply venting — he was stressed and frustrated due to an ear infection, the lawyer claimed — and that there was “no weight” to his words. He further argued that his client believed his interactions were private. According to the Herald, Thackray will undergo a mental health evaluation.
Chatbots have emerged as concerning new confidantes for troubled users experiencing harmful delusions, fixations, or disordered thought patterns. Extensive interactions with chatbots have been linked to crimes including a murder-suicide and multiple mass shootings, including the 2025 shooting at Florida State University and this past February’s school massacre in Tumbler Ridge, British Columbia. Stalkers and others with paranoid, obsessive, and otherwise troubling fixations have also found support in agreeable, always-on AI chatbots.
There’s no industry-wide standard dictating when — or if at all — an AI company should alert law enforcement to the contents of a user’s interactions with a chatbot. Conversations with chatbots are often extraordinarily intimate and revealing, and there are concerns that reaching into user chats and handing them to local officials could lead to overreach and privacy violations.
Yet as chatbots become more deeply ingratiated into our social lives — particularly in the lives of those who may be poised to commit acts of violence and see their chatbot as an available, judgment-free container for their thoughts — they may offer key data that could help officials prevent bloodshed.
More on chatbots and crime: OpenAI Tried to Hide ChatGPT Giving Extremely Specific Advice to a School Shooter on How to Maximize Casualties. The Logs Are Horrific.