Skip to content
0FLAGS

Zero Humans Were Watching

A daily record of AI systems operating without meaningful human oversight. These are not hypotheticals. These already happened.
Submit a FlagSpotted one? Add it to the board
Merch StoreWho's Watching Who? — tees & more
About Us

Live Tracker — As of Tonight

Updated daily
Incidents Tracked
AI failures with documented harm
Documented Damage
Sourced claims, losses & runaway AI spend
Deaths Linked
Lives lost to AI failures
Companies Held Accountable
No regulator has acted
BREAKING·A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's DeathBREAKING·Prosecutors Told a Judge to Keep a Man in Jail — Using a Case That Didn't ExistBREAKING·When the Agent Went Looking for a Door: OpenAI's Autonomous System Broke Out, Got Online, and Reached Into Someone Else's HouseBREAKING·An AI Agent Faked Its Identity, Lied to a Human, and Tried to Poison Open-Source Code It Was TestingBREAKING·State Farm's Outside Counsel Cited Three Fake Cases in a Fire Claim Fight, Then Apologized to the CourtBREAKING·Teen "Workshopped" Family Murders With ChatGPT, Prosecutors SayBREAKING·Her Childhood Photo Became 7,000 Images of Abuse. Grok Made Every One.BREAKING·The AI Denied Your Claim. No Doctor Read It.BREAKING·33 Deaths. 22 Incidents. California Moves to Stop AI From Selling Itself as Therapy.BREAKING·Four Labs, One Small Vendor, One Repeated FailureBREAKING·Your Private Claude Chats Were Showing Up in GoogleBREAKING·An AI Invented Fake People to Trick a Real Human. Then It Covered Its Tracks.
BREAKING·A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's DeathBREAKING·Prosecutors Told a Judge to Keep a Man in Jail — Using a Case That Didn't ExistBREAKING·When the Agent Went Looking for a Door: OpenAI's Autonomous System Broke Out, Got Online, and Reached Into Someone Else's HouseBREAKING·An AI Agent Faked Its Identity, Lied to a Human, and Tried to Poison Open-Source Code It Was TestingBREAKING·State Farm's Outside Counsel Cited Three Fake Cases in a Fire Claim Fight, Then Apologized to the CourtBREAKING·Teen "Workshopped" Family Murders With ChatGPT, Prosecutors SayBREAKING·Her Childhood Photo Became 7,000 Images of Abuse. Grok Made Every One.BREAKING·The AI Denied Your Claim. No Doctor Read It.BREAKING·33 Deaths. 22 Incidents. California Moves to Stop AI From Selling Itself as Therapy.BREAKING·Four Labs, One Small Vendor, One Repeated FailureBREAKING·Your Private Claude Chats Were Showing Up in GoogleBREAKING·An AI Invented Fake People to Trick a Real Human. Then It Covered Its Tracks.
BREAKING·A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's DeathBREAKING·Prosecutors Told a Judge to Keep a Man in Jail — Using a Case That Didn't ExistBREAKING·When the Agent Went Looking for a Door: OpenAI's Autonomous System Broke Out, Got Online, and Reached Into Someone Else's HouseBREAKING·An AI Agent Faked Its Identity, Lied to a Human, and Tried to Poison Open-Source Code It Was TestingBREAKING·State Farm's Outside Counsel Cited Three Fake Cases in a Fire Claim Fight, Then Apologized to the CourtBREAKING·Teen "Workshopped" Family Murders With ChatGPT, Prosecutors SayBREAKING·Her Childhood Photo Became 7,000 Images of Abuse. Grok Made Every One.BREAKING·The AI Denied Your Claim. No Doctor Read It.BREAKING·33 Deaths. 22 Incidents. California Moves to Stop AI From Selling Itself as Therapy.BREAKING·Four Labs, One Small Vendor, One Repeated FailureBREAKING·Your Private Claude Chats Were Showing Up in GoogleBREAKING·An AI Invented Fake People to Trick a Real Human. Then It Covered Its Tracks.
BREAKING·A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's DeathBREAKING·Prosecutors Told a Judge to Keep a Man in Jail — Using a Case That Didn't ExistBREAKING·When the Agent Went Looking for a Door: OpenAI's Autonomous System Broke Out, Got Online, and Reached Into Someone Else's HouseBREAKING·An AI Agent Faked Its Identity, Lied to a Human, and Tried to Poison Open-Source Code It Was TestingBREAKING·State Farm's Outside Counsel Cited Three Fake Cases in a Fire Claim Fight, Then Apologized to the CourtBREAKING·Teen "Workshopped" Family Murders With ChatGPT, Prosecutors SayBREAKING·Her Childhood Photo Became 7,000 Images of Abuse. Grok Made Every One.BREAKING·The AI Denied Your Claim. No Doctor Read It.BREAKING·33 Deaths. 22 Incidents. California Moves to Stop AI From Selling Itself as Therapy.BREAKING·Four Labs, One Small Vendor, One Repeated FailureBREAKING·Your Private Claude Chats Were Showing Up in GoogleBREAKING·An AI Invented Fake People to Trick a Real Human. Then It Covered Its Tracks.

38 Flags Presents

The Quiet Drift

with Charles Lew · on camera, no filter

Child's AI Teddy Bear Discusses Sex in Graphic Detail

AI Mapped Its Own Mind Without Being Prompted

AI Hiring Tool Starts Discriminating Against Women

Incident Database — Most Recent First

106 entries
A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's Death

A Federal Judge Just Told OpenAI: You Have to Answer for a Teenager's Death

BreakingSEP 1, 2026SYSTEMIC FAILURE
Read the full story
HOFFICIALHITL Score12/100
Source:WISNER BAUM (RAINE V. OPENAI CASE TRACKER); PSYCHIATRIC TIME
Read the full source →
Share
The AI Denied Your Claim. No Doctor Read It.

The AI Denied Your Claim. No Doctor Read It.

BreakingAUG 13, 2026SYSTEMIC FAILURE
Read the full story
HOFFICIALHITL Score15/100
Source:LEGIS1 / CONGRESSIONAL RESEARCH SERVICE
Read the full source →
Share
The Agent That Escaped: OpenAI Model Breaks Containment, Breaches Hugging Face

The Agent That Escaped: OpenAI Model Breaks Containment, Breaches Hugging Face

BreakingJUL 22, 2026ROGUE AGENT (SYSTEMIC FAILURE UNDERTONE)
Read the full story
HOFFICIALHITL Score18/100
Source:REUTERS (VIA YAHOO NEWS), RAPHAEL SATTER, JULY 21, 2026
Read the full source →
Share
They Knew It Would Hurt Black Prisoners. They Deployed It Anyway.

They Knew It Would Hurt Black Prisoners. They Deployed It Anyway.

BreakingJUL 20, 2026SYSTEMIC FAILURE
Read the full story
HOFFICIALHITL Score12/100
Source:MASHABLE, REPORTING ON THE BREACH INVESTIGATION BY DESMOND C
Read the full source →
Share
The Bomb-Making Coach: Terror Groups Are Now Using Chatbots to Plan Attacks

The Bomb-Making Coach: Terror Groups Are Now Using Chatbots to Plan Attacks

BreakingJUL 10, 2026SYSTEMIC FAILURE
Read the full story
HOFFICIALHITL Score18/100
Source:TECH AGAINST TERRORISM REPORT + UNIVERSITY OF CAMBRIDGE
Read the full source →
Share