Tucker Carlson wanted to see the \u201cangst-filled\u201d Sam Altman: He wanted to hear him admit he was tormented by the power he holds. After about half an hour of couching his fears with technical language and cautious caveats, the OpenAI CEO finally did.\u00a0<\/p>\n
\u201cI haven\u2019t had a good night\u2019s sleep since ChatGPT launched,\u201d Altman told Carlson. He laughed wryly.\u00a0<\/p>\n
In his wide-ranging interview<\/a> with Tucker Carlson, the OpenAI CEO described the weight of overseeing a technology that hundreds of millions of people<\/a> now use daily. It\u2019s less about the Terminator-esque scenarios or rogue robots. Rather, for Altman, it\u2019s the ordinary, almost invisible tweaks and trade-offs his team makes every day. It\u2019s when the model refuses a question, how it frames an answer, when it decides to push back, and when it lets something pass.<\/p>\n Those small design choices, Altman explained, are replicated billions of times across the globe, shaping how people think and act in ways he can\u2019t fully track.<\/p>\n \u201cWhat I lose sleep over is that very small decisions we make about how a model may behave slightly differently are probably touching hundreds of millions of people,\u201d he said. \u201cThat impact is so big.\u201d<\/p>\n One example that weighs heavily: suicide. Altman noted roughly 15,000 people take their lives each week worldwide, and if 10% of them are ChatGPT users, roughly 1,500 people with suicidal thoughts may have spoken to the system\u2014and then killed themselves anyway. (World Health Organization data confirms about 720,000 people per year worldwide take their own lives).<\/p>\n \u201cWe probably didn\u2019t save their lives,\u201d he admitted. \u201cMaybe we could have said something better. Maybe we could have been more proactive.\u201d<\/p>\n OpenAI was recently sued<\/a> by parents who claim ChatGPT encouraged their 16-year-old son, Adam Raine, to kill himself. Altman told Carlson that case was a \u201ctragedy,\u201d and said the platform is now exploring options where if a minor talks to ChatGPT about suicide seriously, and the system cannot get in touch with their parents, that they would call authorities.<\/p>\n Altman added it wasn\u2019t a \u201cfinal position\u201d of OpenAI\u2019s, and that it would come into tension with user privacy.\u00a0<\/p>\n In countries where assisted suicide is legal such as in Canada or Germany, Altman said he could imagine ChatGPT telling terminally ill, suffering adults suicide was \u201cin their option space.\u201d But ChatGPT shouldn\u2019t be for or against anything at all, he added.\u00a0<\/p>\n That trade-off between freedom and safety runs through all of Altman\u2019s thinking. Broadly, he said adult users should be treated \u201clike adults,\u201d with wide latitude to explore ideas. But there are red lines.\u00a0<\/p>\n \u201cIt\u2019s not in society\u2019s interest for ChatGPT to help people build bioweapons,\u201d he said flatly. For him, the hardest questions are the ones in the gray areas, when curiosity blurs into risk.\u00a0<\/p>\n Carlson pressed him on what moral framework governs those decisions. Altman said the base model reflects \u201cthe collective of humanity, good and bad.\u201d\u00a0<\/p>\n OpenAI then layers on a behavioral code\u2014what he called the \u201cmodel spec<\/a>\u201d\u2014informed by philosophers and ethicists, but ultimately decided by him and the board.\u00a0<\/p>\n \u201cThe person you should hold accountable is me,\u201d Altman said. He stressed his aim isn\u2019t to impose his own beliefs but to reflect a \u201cweighted average of humanity\u2019s moral view.\u201d\u00a0<\/p>\n That, he conceded, is an impossible balance to get perfectly right.<\/p>\n