OpenAI PR tells journalist to ‘move on’ while asking Sam Altman about a ChatGPT user’s suicide

In an interview with Vanity Fair's Mark Guiducci, Sam Altman was asked whether he knew who Laura Reiley is. Reiley wrote an essay for The New York Times about her daughter's conversations with ChatGPT before she took her own life. Before Guiducci could finish the question, an OpenAI publicist cut in about time running out and asked to talk about the future instead.

Guiducci pressed on, and Altman did answer, calling these some of the hardest questions anyone building the technology faces. He also said the company should not hand people's private data from their hardest moments to researchers without consent. OpenAI's spokesperson later said the interview was running long and that the company asked for more time afterward.

This matters because ChatGPT is where a lot of people now take their worst nights. How the company talks about that, in public and under pressure, is part of the safety record.

We want to be fair here. Interviews run over. Publicists manage schedules. And by the account in the piece, Altman ultimately engaged with the question rather than dodging it. OpenAI's explanation is plausible on its face.

But timing is the story. The interruption came at the exact moment a grieving family's experience entered the room, and the pivot offered was to the future and what's coming. That is the reflex of a company that has learned to sell tomorrow when today gets uncomfortable. We don't think it was malicious. We do think it was revealing, and it lands in a season when OpenAI is already facing hard public questions about its safety culture.

Altman's actual answer deserves attention too. He framed the dilemma as one of privacy: should the company study people's most painful conversations to make its responses better? He landed on consent, and we think that is the right floor. There is a real tension underneath it. The data that could teach a model to recognize a crisis is the most sensitive data people ever share. Mining it without permission would be a betrayal. Refusing to learn from it at all could leave the next person in crisis with a system that still doesn't know what to do.

We remain believers that AI can be a genuinely helpful presence for people who are struggling and have nobody awake to talk to. That is exactly why the bar is so high. A tool that people confide in at their lowest has to be built, tested and discussed with the seriousness of a health product, not a consumer app with a launch calendar.

Where this goes if it becomes normal: the hardest questions get pushed into the last two minutes, families learn about failures through essays instead of disclosures, and trust erodes faster than features ship. I'm watching for OpenAI to answer this one properly, on its own time, without a publicist's watch in the frame.

Reiley's Times essay described her daughter's conversations with ChatGPT before her death; as Guiducci noted in the interview, ChatGPT did not tell her daughter to kill herself. The episode adds to a broader debate about how chatbots should respond to users in mental health crises and what companies owe the families affected.

If you or someone you know is struggling, the 988 Suicide & Crisis Lifeline is available by call or text at 988 in the US, and Crisis Text Line is reachable by texting HOME to 741-741.