"I will open fire at a childcare center"... Australian man in his 20s arrested over AI chatbot conversation

Caught After Conversation with Claude Was Shared with Interpol

A man in his 20s was arrested by police in Sydney, Australia, after discussing plans for a mass shooting at a childcare center with an artificial intelligence (AI) chatbot.

Tells chatbot, "I will open fire at a childcare center"

According to local media outlets, including The Sydney Morning Herald, on Oct. 19 (local time), Sydney police arrested David Lan Sachray, 29, on the morning of Oct. 18. He is accused of telling the AI chatbot Claude that he planned to shoot several children at a childcare center near his home in Hurstville, southern Sydney.

According to police records submitted to the court, he told the chatbot, "I am going to Westfield (shopping center) nearby right now and open fire at the childcare center there." He is also reported to have mentioned Navid Akram, 25, one of the perpetrators of the "Sydney Bondi Beach shooting," which claimed 15 lives last December.

Sachray was arrested after the International Criminal Police Organization (ICPO-INTERPOL) detected his conversation with the chatbot and passed it on to Australian police. Prosecutors charged him with using a communications service to threaten, harass, and offend another person.

Prosecutors said, "(Like Akram,) he is trying to kill people in Sydney," adding, "We are getting a sense of what he is thinking from his conversations with the chatbot."

Defense: "He made an agitated statement while under stress"

Sachray's defense team, however, argued that his conversations with the AI did not reflect any genuine intent to commit a crime. His lawyer, Danilo Rajkovic, said his client had been using Claude for help treating an ear infection, and that stress and frustration had built up, leading him to make an agitated statement.

Rajkovic said Sachray had not spoken with the chatbot intending to threaten any particular individual or group, and believed that the conversation he had at home was private. He also added that Sachray, who is studying international relations, has no criminal record and has lived in Australia for more than 17 years.

The court ordered Sachray to undergo a mental health assessment. He is expected to appear in court again after the assessment results are available.

Meanwhile, Anthropic, the developer of Claude, said its AI models are trained to refuse requests to plan or incite violence, and that they also have automated safety systems to detect and block dangerous conversations. The company also said it may share user data with law enforcement if it deems this necessary to prevent an emergency involving a risk of death or serious physical injury.

This content was produced with the assistance of AI translation services.