Reminded every day how lucky we are that Biden and Kamala aren't writing AI policy. Often reminded by Marc Andreessen's clip w/ Joe Rogan.

A team up with Dario and they would think a handful of people should dictate AI behind closed doors and dictate what the world is allowed to have.
Pretty totalitarianism in my view.
It also gives a glimpse into the populace that's gone anti-AI. I think they're wrong. But they're right if the worldview is the one we just saw with Mythos launch.
The man pretends to be a white knight and then plays both sides.
Says AI is too dangerous for open competition, so fearmongers Washington to regulate it so no new entrants.
Says AI is going to destroy jobs at scale no one has ever seen before. Most tech CEOs disagree with him and say it's conflating outcomes (at least near term, obviously labor will adapt and evolve).
But then he goes on to release the strongest model ever. Gate keeps it with the tech oligarchs and makes sure they consolidate power, while the average person effectively is left in the dust.
Anthropic's own written goal: give those players an "asymmetric advantage." Dude is actively rigging the market to kill new founders, small businesses, while empowering the incumbents.
What scares me the most is that he thinks it's for the greater good? A form of effective altruism.
Dude wants to be the AI dictator.
You can't be left to your own whims. So let me be king and I will decide what's best that aligns with my view for a perfect society.
Wants to decide what actions the government can take in wartime usage. His cleared employees get to decide whether the military can even use it in a strike. A private company, vetoing the United States military.
He says fear corporate power. While he builds the biggest corporate-state lock on a technology in history.
Who gets the model. Who gets compute. Who gets security. Who gets an export license. Who gets to deploy.
All of it routes through Anthropic and a dozen approved friends.
This man is building the AI cartel. OPEC of the Magnificent 12. They call it Glasswing.
And then he thinks we are morons with a neutered version of Claude Fable 5 that's literally tampering with workflows deemed a threat and outright blocking any breakthroughs at a consumer level.
A handful of people in San Francisco now decide what you're allowed to learn, build, and know.
They burned books for less. They told women they couldn't read for less.
This is the biggest consolidation of power tech has ever pulled.
Yann LeCun calls it regulatory capture. Says it ends with a few companies owning AI.
David Sacks called Anthropic's playbook regulatory capture built on fear-mongering.
Jeremy Howard and Nathan Lambert watched the Fable launch and called it what it is. Throttling outside builders to widen the gap. Anti-science.
Little Tech sees it. Founders locked out so the incumbents win.
I'm pro capitalism. I'm pro technology. I'm pro every builder this man is trying to lock out.
The answer isn't to trust the king.
Open models. Open weights. Real competition. The frontier in everyone's hands, not behind a velvet rope for a dozen approved friends.
Founders, builders, every lab told to wait your turn. This fight is yours.
Don't let one man decide what humanity is allowed to know or do.
Surprised open-source Chinese models are less totalitarian than what we get from Dario and Sam.