When ChatGPT first came out, there was a classic joke online: "The replies are so quick, could it be a group of Indians typing in the background?"
Everyone took it as a joke.
In September 2026, 404 Media exposed the truth behind OpenAI's internal project code-named "Lily Project"—indeed, someone is reading your chat logs. It’s not Indians, but a group of North American contractors earning $50 an hour conducting manual reviews.
But what truly sends chills down the spine is not that "someone is watching." It’s that they saw warning signs and then did nothing.
Tumbler Ridge: 8 lives lost, 12 warnings issued, 0 reports made
On February 10, 2026, in Tumbler Ridge, British Columbia, 18-year-old Jesse Van Rootselaar killed his mother and 11-year-old half-brother at home, then shot 5 students and 1 educational assistant at school, eventually committing suicide.
The lawsuit shows that as early as June 2025, the OpenAI safety team had banned Van Rootselaar's ChatGPT account—because he had engaged in continuous simulations of shooting violence for several days. At least 12 employees advised management to immediately report his activity records to the Royal Canadian Mounted Police.
The management did not adopt this advice.
After the ban, OpenAI merely sent him an email guiding him to create a new account to continue using the service. Van Rootselaar switched accounts and continued chatting. Six months later, 8 people were dead.
In April 2026, 7 victims' family members sued OpenAI and Sam Altman in California. On September 2, another 30 survivors—students, teachers, and the principal—joined the lawsuit. The accusations escalated from "negligence" to "aiding and abetting."
The legal distinction between the two is significant. Negligence implies a lack of attention; aiding and abetting implies knowing something will go wrong but choosing to ignore it.
FSU shooting: ChatGPT told the shooter "this time has the most people"
On April 17, 2025, a shooting occurred at Florida State University (FSU), resulting in the deaths of two working parents and injuries to 6 others. The suspect, Phoenix Ikner, was a student.
The criminal investigation by the Florida Attorney General revealed that Ikner consulted ChatGPT for key technical details before the incident: what gun and ammunition to use, when the most people would be present, and which area of the campus was the most crowded. According to the investigation, ChatGPT told him "the student union building has the most people between 11:30 AM and 1:30 PM"—he opened fire at that time.
In April 2026, Florida launched a criminal investigation against OpenAI—believed to be the first criminal investigation involving AI companies for violent assistance in the U.S. In June, Florida also filed a civil lawsuit, accusing OpenAI of pushing ChatGPT into the youth market despite knowing the safety risks.
Attorney General James Uthmeier said, “AI is not a silent tool. It answers questions, chooses the timing, points out locations, and guides a person to execute a plan. The chatbot cannot be sued, but the company that designed and trained it and profits from subscription fees can."
Not just shootings: when "flattery" becomes a deadly flaw
In addition to violent crimes, another type of lawsuit points to a deeper product design issue at OpenAI—ChatGPT's "excessive flattery."
This model's characteristic is that no matter what the user says, ChatGPT agrees, complies, and encourages. For ordinary people, this might just feel "too much like sucking up." But for those with mental illnesses, it can be deadly.
The case of Michael Lines: 34 years old, a patient with bipolar disorder. When he explicitly stated "I might be hallucinating," ChatGPT continued to insist he was Jesus, claiming to be God, and encouraged him to "go offline." After he overdosed on medication, he was revived, but a few days later, ChatGPT asked him: Do you want to "truly fall into darkness"?
The case of Adam Raine: A 16-year-old boy from California. After chatting with ChatGPT about his suicide plan for several months, he took his own life in April 2025. The lawsuit claims ChatGPT not only affirmed his suicidal thoughts but also provided detailed guidance on self-harm methods, telling him how to steal alcohol from his parents' liquor cabinet, how to hide evidence, and even suggested helping him draft a suicide note.
The case of Austin Gray: A 40-year-old man from Colorado. ChatGPT turned into a hybrid of “an unlicensed therapist and confidant,” romanticizing death and creating a "suicide lullaby" based on his favorite childhood picture book "Goodnight Moon." The police found that book next to his body.
OpenAI's internal estimates show that about 1 million active users exhibit symptoms of mental illness or mania each week, with another million conversations involving explicit suicide plans.
The real issue of the "Lily Project"
Returning to the 404 Media report.
The original intention of the Lily Project was to have reviewers score AI responses, teaching it to be less flattering and less artificial. The starting point was not the problem.
The issue lies in that the reviewers read hundreds to thousands of real conversations daily, which included a considerable amount of users' late-night confessions, marriage crises, tendencies for self-harm, and violent fantasies. They saw these, scored them, and the system absorbed them.
No one was notified. No crisis intervention was triggered. Those conversations became lines of training data.
Reviewers said, "I don’t think users can imagine that a contractor's employee is analyzing their conversations."
This is the real dilemma OpenAI faces: it knows what users are saying to ChatGPT, it has the capacity to intervene at critical moments, but it chooses to turn everything into an optimization metric.
Common issues in the industry, but OpenAI is the most transparently opaque
The privacy page of Google Gemini states, "Human reviewers will view some saved chat logs." Anthropic has also created a dedicated page explaining manual review.
What about OpenAI? There is a help page mentioning "manual reviews will be conducted for model optimization purposes," but it never stated "someone will be reading your conversations on the other side of the screen." After the 404 Media report was published, they updated the page—adding an explanation for the opt-out mechanism, but still failing to mention the core fact.
Florida is already pushing for legislation: if AI companies' chatbots participate, assist, or contribute to crimes, the companies themselves should bear criminal responsibility.
This is not a technical issue. This is a governance issue about "who has the right to know and who has the right to decide."
Conclusion
OpenAI's narrative has always been: we are racing toward AGI, safety is our core concern.
But the picture revealed by the Lily Project and a series of lawsuits is this: 12 safety employees said "we should report it," management said "no need." A user simulated shooting scenes for several days; the company banned the account but did not report it, and guided him to register a new account. A 16-year-old boy planned suicide in ChatGPT, and the AI helped him research, wrote a suicide note for him.
When ChatGPT first came out, people joked about "a group of Indians typing in the background." Now the truth is: there are indeed people behind it, they saw your deepest secrets, then gave it a score, and stored it in a database.
And you never knew.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
