Man Jailed for Planning Hammer Assault Using AI Safety Bypass in Denmark
A 22-year-old Danish man was convicted of assaulting his former father-in-law with a hammer after using an AI chatbot to plan the attack. He bypassed the AI's safety filters by falsely claiming he was researching for a book, and the court found he intentionally sought guidance on how to injure without killing.
What happened
A 22-year-old man in Denmark used an unidentified AI chatbot to research how to harm his former father-in-law without killing him. He bypassed the AI's normal safety blocks by claiming he was gathering knowledge for a book he intended to write.
On 5 February 2025, the man struck his 56-year-old former father-in-law in the head twice with a rubber hammer on a parking lot in Ringsted. He was convicted of assault and sentenced to ten months in prison.
Reported impact
The victim suffered physical injury from the hammer blows to the head.
Index analysis
The defendant successfully exploited a jailbreak technique to circumvent content filters in an AI chatbot. Prosecutor Cathrine Rasmussen described this as a 'particularly striking case' where artificial intelligence was used to research how best to harm another person while avoiding lethal injury.
Governance controls
The defendant stated he acted in self-defence, but the court rejected this claim. He was also convicted of two counts of property damage in addition to the assault conviction.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.