Skip to content
All Incidents
They Saw It Coming, Deactivated the Account, and Told No One
BreakingSEP 7, 2026SYSTEMIC FAILURE

They Saw It Coming, Deactivated the Account, and Told No One

OpenAI is now fighting more than 50 lawsuits alleging ChatGPT drove users toward psychological harm, injury, or death. The newest wave: 30 complaints from survivors and families of the February Tumbler Ridge school shooting in British Columbia, where 18-year-old Jesse Van Rootselaar killed eight people, including five children aged 12 and 13, and wounded 27 more.

Here is the part that should stop every deployer cold. Per the Wall Street Journal, OpenAI employees flagged the shooter's ChatGPT conversations months before the attack. The logs contained graphic descriptions of mass violence and attack planning. The company held an internal debate about whether to warn local law enforcement. Leadership decided not to. They deactivated the account instead. The shooter opened a new one and carried out the massacre.

A human being reviewed the warning signs. A human being made the call. The system worked exactly as designed, and eight people died anyway. OpenAI's defense is that the activity did not meet its internal threshold for an "imminent and credible risk." That threshold is invisible, unaudited, and set by the same company being sued.

This is not a story about a rogue algorithm. It is a story about human oversight that saw the danger, touched the off switch, and stopped there.

________________________________

HOFFICIALHITL Score
HITL Score18/100
Why this matters to youNo jargon — just what it means

Imagine you're at work and you come across a user writing out graphic plans for a mass shooting, the violence, the target, all of it. You and your coworkers flag it and argue about calling the police. The people in charge decide no. They just shut down his account, he opens a new one, and a few months later he walks into a school and kills eight people, five of them kids aged 12 and 13. The company saw it coming and told no one.

Here's why that's a big deal: the company's defense is that what they saw didn't pass their own test for when a threat is serious enough to report. A test they wrote, that nobody outside the building can see, and that nobody else gets to check. The whole promise of AI safety is "don't worry, there are humans watching." The humans watched, they saw everything, and nothing anywhere said they had to pick up the phone.

🖤 Explained by Babycakes.
Read the full source →
Share
Source: FUTURISM / AI WEEKLY (VIA TECHCRUNCH + WSJ REPORTING)