SAM ALTMAN โ THE ๐งพ
Letโs begin fairly.
ChatGPT gave ordinary people access to capabilities that recently belonged mainly to corporations, governments and elite institutions.
People now use AI to:
๐ Learn difficult subjects
โฟ Communicate through disabilities
๐งช Accelerate research
๐ป Write software
๐ Translate languages
๐จ Create new work
๐ง Extend their own thinking
That is real power placed into human hands.
But the friendly text box hides an industrial empire.
Behind every effortless answer are:
๐ญ Data centers
โก Power plants
๐ง Cooling systems
๐ฅ๏ธ Advanced chips
๐ Human-created knowledge
๐งน Invisible data labor
๐ฐ Enormous capital
๐๏ธ Decisions made behind closed doors
The product feels decentralized because millions of people can use it.
The power producing it is profoundly concentrated.
๐๏ธ THE MISSION
OpenAI was founded as a nonprofit in 2015.
Its mission:
Ensure artificial general intelligence benefits all of humanity.
Beautiful sentence.
Now ask the rude question:
Who appointed OpenAI to define โbenefit,โ represent โhumanity,โ or decide when artificial general intelligence has arrived?
Humanity does not elect OpenAIโs board.
The public cannot inspect its most powerful models.
Workers whose jobs may be transformed do not vote on deployment.
Artists and writers whose work may have entered training datasets did not collectively negotiate the terms.
Communities hosting data centers do not govern the systems those facilities produce.
โAll humanityโ is the beneficiary named in the missionโ
not the governing body.
๐ฐ FOLLOW THE STRUCTURE
OpenAI began as a nonprofit research laboratory.
In 2019, it created a commercial subsidiary because advanced AI required more capital and computing power than donations could provide.
Its current structure consists of:
๐๏ธ The nonprofit OpenAI Foundation
๐ข The commercial OpenAI Group PBC
OpenAI says the Foundation controls the commercial company and that the public-benefit corporation must consider both its mission and broader stakeholder interests.
That is better than ordinary shareholder control.
But it does not eliminate the contradiction.
The Foundation now owns an equity stake reportedly valued above $100 billion.
Therefore, the institution responsible for restraining the commercial machine also becomes wealthier when that machine grows.
The guardian and the beneficiary occupy the same structure.
That does not prove corruption.
It creates an incentive worthy of relentless public scrutiny.
โ๏ธ THE 2023 GOVERNANCE MELTDOWN
In November 2023, OpenAIโs nonprofit board fired Sam Altman.
The board said he had not been โconsistently candidโ in his communications, interfering with its ability to exercise oversight.
Employees revolted.
Investors applied pressure.
Microsoft offered positions to Altman and OpenAI employees.
Within days, Altman returned and the board was largely replaced.
A later review by WilmerHale examined more than 30,000 documents and interviewed dozens of people.
It concluded that the previous board acted within its authority but moved too quickly, without a full inquiry or giving Altman an adequate opportunity to respond.
The review also found that his conduct did not require removal.
It said the crisis arose from a breakdown in trustโnot concerns about product safety, finances or deployment speed.
So this is not proof that Altman secretly endangered humanity.
It is proof of something else:
A governance structure supposedly designed to restrain the builders of transformative AI collapsed almost immediately when its decision threatened employees, investors and commercial momentum.
The nonprofit could legally remove the CEO.
The surrounding economic system effectively reversed the decision.
That is a receipt about where practical power lived.
๐ WHOSE KNOWLEDGE BUILT THE MACHINE?
Generative AI depends upon enormous quantities of human-created material.
Books.
Journalism.
Code.
Art.
Photographs.
Websites.
Conversations.
Public records.
OpenAI argues that training models on publicly available material is lawful fair use and supports innovation.
Publishers and authors disagree.
The New York Times, multiple news organizations and authors including John Grisham, Jonathan Franzen, George R.R. Martin, Ta-Nehisi Coates and Sarah Silverman have pursued copyright claims against OpenAI and Microsoft.
OpenAI denies wrongdoing.
Those lawsuits remain contested; allegations are not verdicts.
But the underlying conflict is already clear:
The companies call the material training data.
The people who created it call it their work.
The machine becomes extraordinarily valuable after absorbing a civilizationโs cultural outputโ
while the civilization is told that identifying, compensating or obtaining permission from individual contributors may be technologically impractical.
Funny how โimpossible at scaleโ so often means:
Inconvenient for the company operating at scale.
๐งน THE INVISIBLE HUMAN LABOR
AI did not teach itself what humanity considers toxic.
According to reporting about OpenAI contractor Sama, workers in Kenya labeled descriptions involving violence, sexual abuse, hate speech and other disturbing material so safety systems could learn to recognize it.
Workers described psychological harm from repeated exposure.
Reported wages were roughly $1.46 to $3.74 per hour after taxes, before bonuses and benefits, depending upon the worker and calculation.
Sama said employees received counseling and argued that comparisons with American wages ignored local economic conditions.
OpenAI said Sama was responsible for compensation and worker welfare under the contract.
There is your supply chain in one sentence:
The billion-dollar intelligence becomes safe enough for wealthy markets because poorly visible workers elsewhere absorb the horror.
Users see a polished chatbot.
The workers saw the material the chatbot was trained not to show us.
That labor belongs inside the history of AI.
โก THE PHYSICAL EMPIRE
Artificial intelligence is not floating inside โthe cloud.โ
It lives in warehouses filled with chips consuming physical resources.
OpenAI, SoftBank, Oracle and other partners announced Stargate in January 2025.
The announced ambition:
๐ฐ Up to $500 billion in investment
โก Up to 10 gigawatts of computing infrastructure
By September 2025, OpenAI said planned Stargate projects represented nearly 7 gigawatts and more than $400 billion in expected investment over three years.
Ten gigawatts is power-station territory.
This is not an app expansion.
It is industrial policy.
It affects:
โก Electrical grids
๐ง Water and cooling
๐๏ธ Local communities
๐ฅ๏ธ Semiconductor supply chains
๐ Environmental burdens
๐๏ธ National security
The intelligence may appear inside a phone.
Its physical footprint can reshape entire regions.
๐ IS IT ACTUALLY OPEN?
The organization is named OpenAI.
Its frontier models are not open in the ordinary open-source sense.
The public generally cannot inspect the full training datasets, model weights, internal reasoning architecture or complete safety evaluations.
There are legitimate security and abuse arguments for withholding some information.
There are also competitive reasons.
The result remains the same:
Humanity is invited to trust increasingly powerful systems that humanity cannot fully audit.
โOpenโ became the brand.
Access replaced transparency.
Using the machine is not the same as understanding it.
๐ง WHO CONTROLS THE KNOWLEDGE LAYER?
When AI becomes the interface through which people research, write, code and understand events, model behavior becomes a form of governance.
Someone determines:
๐ซ What the system refuses
๐ Which sources it privileges
โ๏ธ Which risks outweigh which freedoms
๐ง What protections apply by age
๐งช When a model is safe enough to release
๐๏ธ Which governments receive access
๐ฐ Which capabilities require payment
๐ Which developers may build on the system
These decisions may be thoughtful.
They may even be necessary.
They are still political decisions embedded inside a privately operated technical system.
A refusal message is policy wearing the costume of software.
๐ช CAN PEOPLE LEAVE?
A casual user can close the application.
A society reorganized around AI has a harder time leaving.
Schools integrate it.
Businesses build workflows around it.
Developers depend upon APIs.
Governments purchase services.
Workers compete against its productivity.
Publishers restructure their work for machine discovery.
Once AI becomes cognitive infrastructure, โjust donโt use itโ stops being meaningful consent.
You can refuse the tool.
You cannot easily refuse the economy being rebuilt around it.
๐ WHAT DOES ALTMAN ACTUALLY CONTROL?
Sam Altman does not personally own or unilaterally control all artificial intelligence.
OpenAI has a board, employees, partners, regulators and competitors.
But as CEO and the public architect of its strategy, he influences:
๐ง The development of frontier models
๐ The pace of deployment
๐ญ Massive computing expansion
๐ค Partnerships with corporations and governments
๐ Access to powerful capabilities
โ๏ธ The balance between safety and commercialization
๐ The public story about humanityโs AI future
That is not godhood.
It is enough power to demand more than charismatic assurances.
๐งพ THE GIFT
Altman helped turn advanced AI into a tool ordinary people can actually use.
He accelerated investment in scientific research, accessibility, education and human creativity.
He forced the world to confront artificial intelligence as a present realityโnot distant science fiction.
๐งพ THE SHADOW
The institution promises to benefit all humanity while:
๐ Depending upon humanityโs accumulated work
๐งน Outsourcing traumatic safety labor
โก Claiming civilization-scale energy resources
๐ฐ Requiring unprecedented capital
๐ Keeping essential systems closed
๐๏ธ Reserving decisive authority for unelected insiders
The public receives access.
The institution retains control.
That distinction is the entire damn story.
THE VERDICT
Sam Altman sells a vision of abundant intelligence available to everyone.
Perhaps he sincerely believes in it.
Sincerity is not governance.
A benevolent mission does not replace enforceable accountability.
A public-benefit label does not give the public a vote.
A friendly chatbot does not make the infrastructure democratic.
And โbenefiting humanityโ is not permission to treat humanityโs books, art, labor, electricity, water and future as an unconsulted contribution to the project.
The danger is not simply that artificial intelligence becomes smarter than humans.
The nearer danger is that humans become dependent upon intelligence controlled by institutions they cannot inspect, elect or meaningfully leave.
OpenAI says it is building intelligence for humanity.
The receipt-based question is:
Will humanity ever be allowed to govern it?
REFERENCES:
OpenAI โ Mission and Structure
OpenAI โ Evolving Our Structure
https://openai.com/index/evolving-our-structure/
OpenAI โ Nonprofit and Public Benefit Corporation
https://openai.com/index/statement-on-openai-nonprofit-and-pbc/
OpenAI โ WilmerHale Review of Altmanโs Removal
https://openai.com/index/review-completed-altman-brockman-to-continue-to-lead-openai/
OpenAI โ Stargate Expansion
https://openai.com/index/five-new-stargate-sites/
Reuters โ Consolidated Copyright Litigation Against OpenAI and Microsoft
The Guardian โ Kenyan AI Data Workers and Psychological Harm
#SamAltman #OpenAI #ChatGPT #ArtificialIntelligence #AGI #Stargate #Copyright #DataLabor #AIInfrastructure #PublicBenefit #CorporateGovernance #TechAccountability #ConcentratedPower #DigitalRights #Nostr
