Public transcript

Sam Altman’s Dystopian Vision to Replace God With AI

By: Tucker Carlson Network
Length: 18 min
Language: English
Text from: the video's captions
Transcribed: 12 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

In this Tucker Carlson Network interview, Sam Altman discusses his spiritual views, the concentration of power around AI, and how OpenAI sets the moral framework for ChatGPT. Carlson argues the technology functions like a religion that should be transparent about what it stands for, while Altman says the model spec documents the rules and that he, not his team, should be held accountable. Altman says he worries most about unknown societal-scale effects, giving the example of people adopting ChatGPT's writing style.

Key points

Sam Altman identifies as Jewish and believes there is something bigger going on than can be explained by physics, though he has never felt communication from a higher force.
Altman says he used to worry about AI concentrating power in a few hands, but now believes it will broadly up-level everyone who embraces the technology.
OpenAI trains its base model on the collective experience and knowledge of humanity, then aligns it using a written model spec.
Altman says they consulted hundreds of moral philosophers when creating the model spec, and that he personally should be held accountable for those decisions.
He says ChatGPT should reflect a weighted average of humanity's moral views rather than his own, and that users should be allowed to hold very different moral views.
Altman worries most about unknown unknowns, such as societal-scale behavior changes from millions of people talking to the same model, like people picking up ChatGPT's writing style and overuse of M dashes.
Carlson argues the technology is effectively a religion that should transparently state what it stands for; Altman responds that the model spec exists precisely so people can see how the model is intended to behave.

Questions it answers

00:00Spiritual views

What does Sam Altman believe about God and higher powers?

Altman says he is Jewish with a fairly traditional view of the world, believes there is something bigger going on than physics can explain, and does not think existence feels like a spontaneous accident. He has never felt communication from a force beyond the material.

He identifies as Jewish but is not a Biblical literalist.
He believes in a mystery beyond his comprehension but no personal communication from it.

01:24Power concentration

Does Altman think AI will concentrate power in a few hands?

He used to worry about that, but now believes AI will broadly up-level everyone who embraces it, with billions of users becoming more capable. He says the trajectory could shift again and they would have to adapt.

Broad distribution of capability worries him much less than a small group gaining power.
He acknowledges the future trajectory is uncertain.

02:47Moral framework of ChatGPT

How does OpenAI decide what is right or wrong for ChatGPT?

The base model learns from the collective experience and knowledge of humanity, then is aligned using a written model spec that defines what it will answer or refuse. They consulted hundreds of moral philosophers and invite world input through a debate process.

The model spec lets people tell whether behavior is a bug or intended.
Users have convinced OpenAI to change decisions they thought were easy calls.

05:29Accountability

Who is accountable for ChatGPT's moral decisions?

Altman says he is the person to hold accountable, since he is the public face and can overrule decisions, along with the board. He says he doesn't sleep well and worries most about small behavioral decisions affecting hundreds of millions of users.

He turns 40 this spring and feels the weight of these decisions.
He worries more about small decisions at scale than the big moral calls.

09:43Reflecting humanity

Whose moral views should ChatGPT reflect?

Altman says ChatGPT should reflect a weighted average of humanity's moral view, not his own, and that individual users should be allowed to hold very different moral views without the AI calling them wrong or immoral.

He says there are things ChatGPT allows that he personally disagrees with.
He believes users should be treated like adults with broad freedom.

13:13Downsides and unknown unknowns

What downsides of AI does Altman worry about?

He worries most about unknown unknowns rather than known risks like bioweapons, which the industry is actively mitigating. He gives the example of real people picking up ChatGPT's writing style, including overuse of M dashes, as evidence of societal-scale behavior change.

Known risks like bio are being worked on across the industry.
Millions talking to the same model can shift behavior in unpredictable ways.

14:34AI as religion

Why does Carlson call this technology a religion, and how does Altman respond?

Carlson argues it is something more powerful than people that we look to for guidance, and that unlike religions it doesn't transparently state what it stands for, guiding people stealthily. Altman responds that the model spec exists precisely so people can see how the model is intended to behave, and expects it to become very long and complicated.

Carlson says the unsettling part is not knowing what the technology stands for.
Altman says the spec will vary by country and law and grow over time.

Notes

Background and Beliefs

Sam Altman describes himself as Jewish with a fairly traditional worldview, though not a literalist about the Bible.
He believes there is "something bigger going on" than can be explained by physics, but says he has never felt communication from a force beyond the material.
He says his moral framework comes mainly from his upbringing: family, community, school, and religion.

Concentration of Power

Altman says he used to worry a lot about AI concentrating power in one or a handful of people or companies.
He now sees AI as a huge up-leveling where everybody who embraces the technology becomes more powerful, with billions of people using it.
He acknowledges the trajectory could shift again and they would have to adapt.

Training and the Model Spec

One early framing: training the model to be like a single human versus training it to be like the collective of all humanity, learning everything and all perspectives.
The base model is trained broadly, then aligned using a written model spec that defines what the model will answer or refuse.
OpenAI consulted hundreds of moral philosophers and ethics-of-technology experts, but ultimately had to make decisions themselves.
Altman says he is the person who should be held accountable, since he can overrule decisions, along with the board.

User Freedom and Limits

Altman's principle is to treat adult users like adults, with strong privacy and individual freedom guarantees.
He draws absolute bounds where society's interest conflicts with user freedom, e.g., ChatGPT should not teach people how to make bioweapons.
He says ChatGPT should reflect a weighted average of humanity's moral view, not his personal views, and users should be allowed to hold very different moral views.

Risks and Unknown Unknowns

Known risks like AI assisting biological weapons are being actively mitigated across the industry.
He worries most about unknown unknowns, such as people adopting the model's writing style, including overuse of M dashes, showing societal-scale behavior change.

Carlson's Religion Argument

Carlson argues the technology is a religion: something more powerful than people that we look to for guidance.
He says religions are transparent about what they stand for, while AI guides people stealthily toward conclusions.
Altman responds that the model spec exists precisely so people can see how the model is intended to behave, and expects it to grow longer and more complicated over time.

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.