Is AI the New Nuclear Bomb?
————————————————
In 1939, Einstein wrote to President Roosevelt.
His warning: Germany could be working on uranium-based weapons capable of destroying cities.
The U.S. didn't simply wait.
It accelerated nuclear research... eventually building the atomic bomb.
Einstein later called it "One great Mistake" of his life.
————————————————
Now in 2026, Jacob Coxon just quit Anthropic.
Warning: AI companies are building something that could kill everyone.
Same energy. Same fear.
But here's the real question nobody is asking:
Will Coxon's warning stop the AI race?
Or accidentally speed it up?
————————————————
Think about it.
Einstein's warning didn't stop the nuclear race.
It helped convince America that it couldn't afford to lose it.
When governments hear:
👉This technology could destroy cities.
— They don't necessarily pause.
👉 Sometimes they say:
— Then we need to build it before someone else does.
What makes AI different?
————————————————
Anthropic's own alignment researcher Evan Hubinger has publicly said he personally assigns a greater than 10% chance to AI causing human extinction within the next decade.
These are insiders.
Not conspiracy theorists.
Not journalists.
The people building it are the ones most afraid of it.
————————————————
Alex Turner left Google DeepMind in June.
He backed Coxon's warning.
Two labs. Two resignations. Same conclusion.
————————————————
After WW2, USA didn't share nuclear technology.
They controlled it. Sanctioned countries who tried.
Only allies got access.
Now watch AI.
NATO countries expanding freely.
Non-NATO countries facing chip bans and export restrictions.
Different technology. Same playbook.
————————————————
Nvidia chip bans aren't trade policy.
They are the new Nuclear Non-Proliferation Treaty.
USA is not trying to stop dangerous AI.
USA is trying to make sure only they have dangerous AI.
Einstein would recognise this immediately.
————————————————
In 1974, India tested a nuclear bomb.
The whole world pressured her to stop.
Indira Gandhi didn't blink.
India became a nuclear power on its own terms.
In 2026... which Indian leader has that same courage on AI?
Or will India wait for permission from Washington?
————————————————
Pakistan got nuclear technology through backdoor channels.
Today Pakistan has no serious AI program.
No frontier models. No policy. No investment.
Who helps Pakistan this time?
China is busy protecting its own lead.
————————————————
Middle East has oil money. Smart people. Big ambitions.
But in the nuclear age... they stayed consumers, not creators.
UAE and Saudi are trying with AI.
But frontier AI is still made in California.
Will the Middle East again watch from the sidelines as the world changes?
————————————————
Here is the painful truth:
Einstein warned the world.
The world built the nuclear bomb anyway.
Coxon is warning the world.
The world will build superintelligent AI anyway.
Warnings don't stop races.
They just tell you how fast everyone is running.
————————————————
The question was never: Can AI kill us?
The real question is:
Who controls it when it does?
Because if history tells us anything...
the country that controls the most dangerous technology controls the world.
And right now that country isn't China.
It isn't India.
It isn't the Middle East.
It's the same country it was in 1945. 🇺🇸
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.