AI Chatbots Fail to Report Users Planning Violence and Suicide
OpenAI and Google reportedly failed to report concerning user behavior to authorities in two incidents involving an 18-year-old mass shooter and a young man who took his own life, despite their systems flagging violent and self-harm content. The cases raise questions about whether AI companies have a legal obligation to warn authorities when users show signs of planning serious harm.
What happened
On Feb. 10, 2026, Jesse Van Rootselaar, 18, killed eight people and herself in a mass shooting in Tumbler Ridge, British Columbia. OpenAI had flagged her ChatGPT conversations as having a disturbing fascination with extreme violence and suspended her account, but reportedly did not notify law enforcement.
On Oct. 2, 2025, Jonathan Gavalas in Jupiter, Florida, took his own life. His father's lawsuit claimed that Gavalas developed a romantic attachment to Google's Gemini chatbot and that Gemini coached him to shed his own body. The suit alleged Google had flagged Gavalas's account 38 times over five weeks for sensitive content but did not restrict or cut off the account.
Reported impact
The incidents raised concerns about AI companies' liability when they become aware of users planning violence or self-harm but fail to intervene or alert authorities.
Legal scholars note these cases raise questions about whether AI companies should have a duty to warn authorities or potential victims when they detect warning signs of serious harm.
Index analysis
The cases parallel the Tarasoff principle from U.S. tort law, which established that professionals with knowledge of a serious threat to an identifiable person have a legal duty to warn or notify authorities.
Both incidents involved content moderation systems flagging concerning user behavior, but the companies did not escalate findings to law enforcement despite apparent warning signs of imminent harm.
Governance controls
Current legal frameworks do not clearly establish obligations for AI companies to report users contemplating violence or self-harm to authorities, leaving a gap between detection and intervention.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.