Public transcript

FULL REMARKS: OpenAI CEO Sam Altman Warns UN of AI Risks, Calls for Human Control | AI15

By: DWS News
Length: 8 min
Language: English
Text from: the video's captions
Transcribed: 13 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Sam Altman, CEO of OpenAI, addresses the UN Security Council on the risks and promise of AI. He identifies two major risks—losing control of the future to AI and concentrating too much power in too few hands—and lays out three principles for a safe path forward. He calls for international cooperation, common standards for frontier AI, and democratic oversight of the technology's development.

Key points

Altman says AI should be built for people, not simply to accelerate machines for their own sake.
He warns of two ways AI progress could go badly: losing human control of the future, and concentrating too much power in too few hands.
He states that no level of catastrophic risk—10%, 1%, or 0.1%—is remotely acceptable, and models must be kept under human control.
OpenAI has unilaterally slowed down in the past and will do so in the future rather than race rashly.
He outlines three principles: keeping people at the center of AI decision-making, ensuring benefits are by and for people, and empowering people individually.
He cites rapid progress in math, including a model solving the Navier-Stokes equations, a millennium prize problem.
He argues the most important AI decisions cannot be made by labs in San Francisco alone and must be shaped through democratic processes and governments.
He calls for complementary national and international frontier AI standards, incident reporting protocols, and secure channels to share vulnerabilities and threats.
He frames the moment as a crossroads between concentrated, opaque power and a democratic future where AI works for people.

Questions it answers

00:00Promise and anxiety of AI

How does Altman characterize the current moment for AI?

He says the moment is complicated, with both tremendous potential and very understandable anxiety. AI can help discover new knowledge, medicines, better health care, and stronger education, but as systems become more capable and autonomous they could move faster than our institutions or concentrate power.

AI should be built for people, not to turn the crank of machines faster for their own sake.
AI can be a defensive tool against the increasing complexity of life.

01:24Risk one: losing control

What is the first way Altman says AI progress could go badly?

We could lose control of the future to AI, with systems moving so fast that people can no longer follow what's happening or intervene. He says no level of catastrophic risk is acceptable and we should not train models we cannot strongly argue will stay under human control.

Recursive self-improvement could accelerate AI progress rapidly.
OpenAI has unilaterally slowed down in the past and will do so again.
Beating companies in a race is not a reason to make rash decisions.

02:41Risk two: concentrated power

What is the second major risk Altman identifies?

AI systems could concentrate too much power in too few hands, letting one person, company, or country impose their worldview on everyone else. He says we must reject the logic that only one actor can be trusted with the technology.

A belief that only one actor can be trusted can justify almost anything.
Altman believes there is a pragmatic path through both risks.

03:22Three guiding principles

What principles does OpenAI follow for AI development?

First, people must remain at the center of AI decision-making, with alignment ensuring systems stay under human control and reflect human values. Second, the benefits of scientific progress and economic growth must be by people and for people. Third, the technology must empower people individually.

We must not automate human judgment or human values.
The right tools make people more capable rather than replacing human agency.
OpenAI sees itself as a builder, not the hero of the story.

04:43Math progress and discovery

What progress does Altman cite as evidence of AI's capability?

He traces rapid math progress: models were okay at grade school math three summers ago, reached a gold medal at the most prestigious international math competition last summer, and recently solved the Navier-Stokes equations, a millennium prize problem used in aircraft design, weather forecasting, and blood flow studies.

Math progress is aimed at empowering people to discover new knowledge.
The goal is improving health care, raising living standards, and expanding potential.

05:59Democratic oversight and standards

How should decisions about AI be made and coordinated internationally?

Altman says companies must not substitute for the democratic process, and decisions cannot be made by labs in San Francisco alone. He calls for complementary national and international frontier AI standards, incident reporting protocols, and secure channels to share vulnerabilities and threats.

Standards should cover measuring capabilities, assessing risks, and preserving human oversight.
Standards must not lock in incumbents or favor one business model.
Each government should decide how to incorporate standards into its own legal system.

07:58A crossroads

What choice does Altman say the world faces?

He describes two paths: one where power is concentrated, decisions are opaque, and people feel the technology is being done to them; another where AI works for people because democratic institutions help decide and guide its development, giving individuals and communities greater power to live as they choose.

Altman says OpenAI does not have all the answers and is open to better ideas.
He calls the second path a future of realized potential, excitement, and wonder.

Notes

Context and Framing

Sam Altman addresses the UN Security Council, convened by France, on the risks and potential of AI.
He describes the moment as complicated: tremendous potential alongside very understandable anxiety.
AI can help discover new knowledge, medicines, better health care, stronger education, and more productive economies.

Two Ways AI Progress Could Go Badly

Losing control to AI: As systems become more capable and autonomous, they could move faster than our institutions or make decisions people no longer understand or control. This is especially concerning with recursive self-improvement.
Altman says no level of catastrophic risk—10%, 1%, or 0.1%—is remotely acceptable, and we should not train models we cannot make an extremely strong case will stay under human control.
Concentration of power: No one person, company, or country should use the most powerful AI models to impose their worldview on everyone else.

OpenAI's Three Principles

People must remain at the center of AI decision-making; alignment means systems reliably remain under human control and reflect human values.
The benefits of scientific progress and economic growth must be by people and for people.
The technology must empower people individually; OpenAI is a builder but "not the heroes of this story."

Progress and Democratic Oversight

Altman cites math progress: grade school level three summers ago, gold medal at the most prestigious international math competition last summer, and a model solving the Navier-Stokes equations, a millennium prize problem.
He stresses that companies must not substitute for the democratic process; decisions cannot be made by labs in San Francisco alone.

International Cooperation

He calls for complementary national and international frontier AI standards: measuring capabilities, assessing risks, verifying safeguards, and preserving human oversight.
He urges accurate and speedy incident reporting and secure channels among governments, critical infrastructure operators, and technical experts.
Standards should not lock in incumbents and must support open and closed model developers, new entrants, and established labs.