Why is Sam Altman losing sleep? OpenAI CEO addresses controversies in interview
Openai CEO Sam Altman and Advanced Micro Devices CEO Lisa Su, Senate Trade, Science and Transportation Committee during the hearing of “AI race: Strengthening US capabilities in US calculation and innovation” Hart building on May 8, 2025.
Tom Williams | CQ-Rluk Call, Inc. | Getty Images
While sweeping report Last week, Openai CEO Sam Altman discussed numerous moral and ethical questions about the company and the popular Chatgpt AI model.
“Look, I don’t sleep so well at night. I have a lot of weight, but it’s probably nothing but a hundreds of millions of people talking to our model every day,” Altman said in an almost one hour interview with the former Fox News presenter Tucker Carlson.
“Actually, I’m not worried about making big moral decisions wrong,” Altman said, “Maybe we’ll understand them wrong.”
On the contrary, he said he lost the most sleep on “very small decisions” on model behaviors that may have great reflections.
These decisions tend to gather around the ethics that informs Chatgpt and do not answer which questions of Chatbot. It is a draft of these moral and ethical dilemmas, which seems to keep Altman awake at night.
How does Chatgpt handle suicide?
According to Altman, the most difficult problem of the company in recent times is how Chatgpt approaches suicide in the light of a lawsuit. A family accusing Chatbot for the suicide of his young sons.
CEO said many of the thousands of people who committed suicide every week. Talk to chatgpt in the beginning.
“They probably talked [suicide]And we probably didn’t save their lives, “Altman said sincerely.” Maybe we could say something better. Maybe we could be more proactive. Maybe we could give a little better advice, hey, you need to get this help. ”
Last month, Adam Raine’s parents filed a product obligation against Openai and wrong death after his sons died with suicide at the age of 16. In the case, the family said “Chatgpt helps Adam’s discovering suicide methods”.
Shortly after, Blog post Openai, titled “Helping people when they need the most”, planned to deal with the deficiencies of Chatgpt while dealing with “sensitive situations” and said he would continue to improve technology to protect the most vulnerable people.
How to determine the ethics of chatgpt?
Another major issue that emerged in the sitting interview was ethical and ethics that informed ChatGPT and its commissioners.
Altman described Chatgpt’s basic model as trained in collective experience, knowledge and learning, while Openai should later align Chatbot’s certain behaviors and decide what questions he would not answer.
“This is a really hard problem. Now we have a lot of users and they come from very different life perspectives … But in general, I was a pleasant surprise from the model’s ability to learn and apply a moral framework.”
When the decision was pressed on how to decide certain model specifications, Altman said that the company consulted “hundreds of moral philosophers and people who think about technology and system ethics”.
An example of a model specific is that Chatgpt avoids answering questions about how to make biological weapons if desired by users.
“There are clear examples that society has an interest in user freedom with a significant tension,” Altman said, “The company” will not take everything right and at the same time will not need the input of the world. “
How special is chatgpt?
Another great debate was the concept of user privacy about Chatbots, and Carlson argued that the productive AI could be used for “totalitarian control”.
In response, Altman said that a part of a policy he pushed in Washington was “AI privilege”, which expresses the idea that something that a user says to a chat boat should be completely hidden.
“When you talk to a doctor about your health or a lawyer about a lawyer, the government cannot get this information, right? … I think we should have the same concept for AI.”
According to Altman, users will allow users to consult AI Chatbots about their medical history and legal problems, as well as other things. Currently, US officials added to the company that he could court for user data.
“I think the government feels optimistic because we can make it understand its importance.”
Will Chatgpt be used in military operations?
Carlson asked if Chatgpt would be used by the army to damage people by the army, Altman did not respond directly.
“I don’t know the way people in the army use chatgpt today … But I suspect that there are many people who speak for advice with chatgpt in the army.”
Later, he added that he was not sure of “exactly how to feel”.
Openai was one of the AI companies. The US Department of Defense received a $ 200 million contract to put productive artificial intelligence for the US army. Firm He said in a blog post It will enable the US government to access special AI models for national security, support and product roadmap information.
How strong is Openai?
In his interview, Carlson estimated that Sam Altman, a productive AI and extension in his current orbit, could collect enough power to call Chatgpt to Chatgpt.
In response, Altman said he was very worried about the power concentration that may arise from the productive AI, but now he believes that all AI will result in a “great leveling”.
“What is currently using tons of people chatgpt and other chatbotes, and they are all more talented. They all do more.
However, CEO said that AI thinks that he will eliminate many existing work, especially in the short term.
