Me on Claude Lane after getting turned away from the hackathon. The sign was too good to pass up.
Down the block from Corgi Cafe… Hudson pointing at the CLAUDE sign. The universe writes better punchlines than I ever could.
It started on Discord at 8 AM.
I saw the post. Claude’s Builders Day. A hackathon in San Francisco. Same day. I looked at Hudson and said, “Get in the car.”
Two hours later we were pulling into the city. I had no entry. I tried to submit one on the drive, but they did not accept it. I showed up anyway, hoping to talk our way in. They turned us away at the door.
Hudson was disappointed. I was frustrated. But we were already in San Francisco, and I am not the kind of person who lets a drive like that go to waste.
So I did the next best thing.
I was already in San Francisco with Hudson. I thought about where to go. Claude Lane. A real street in the Financial District, right by the Anthropic office. The sign was too good to pass up.
So we went.
Claude Lane is a real street in San Francisco.
Tucked between Bush Street and Maiden Lane in the Financial District. A small service alley with a coffee shop called Corgi Cafe at the very end. Hudson pointed at the “CLAUDE” sign above the storefront and laughed.
We sat down. Hudson pulled out his book. I pulled out Dario Amodei’s “The Adolescence of Technology” on my phone and started reading.
The street is named Claude. The coffee is $7. The essay is about whether humanity can survive its own technological adolescence. I am sitting there with my 11-year-old son, who I am building a whole platform to teach Claude Code to, on a street named after the very thing I am teaching him.
You cannot make this up.
Fable got pulled out of Escalation two days before this.
The US government issued a directive on June 12. Fable 5 and Mythos 5 suspended for all customers. National security export control. Anthropic received the order at 5:21 PM Eastern and complied. A frontier model. Shut down by government order. Not a hypothetical. Not a thought experiment. Happening right now, while Hudson and I sat on Claude Lane.
This is the core of what Dario’s essay is really about. Safety is not a technical problem. It is a political one. The question is not whether we can build aligned AI. The question is whether our institutions have the maturity to handle it.
I am not sure Cloudflare can navigate this with the powers that be.
I say that as someone who works there. Principal SE. Enterprise. Cloudflare sits at a different intersection of the same problem. When the government asks you to do something, you have to decide. Every time. The answer is not always clear.
I am not sure Anthropic can navigate it either.
They got a directive and they complied. Dario is out here writing essays asking for measured, surgical regulation. But regulation does not stay surgical. It grows. The pendulum swings. Right now it is swinging toward control, not permission. And both of these companies I love are going to have to figure out how to operate in that new reality.
If I had to bet on anybody, it would be the seven co-founders and the team there.
Dario, Daniela, Jared, Tom, Chris, James, and the rest. I have watched them be consistent through two pendulum swings. Panic in 2023-2024. Dismissal in 2025-2026. They did not change their message through either one. They kept building. They kept publishing system cards that run hundreds of pages. They kept spending 5% of inference costs on bioweapons classifiers when nobody was watching and nobody would have noticed if they stopped.
Dario’s essay lays out five categories of risk: autonomy risks where AI goes rogue, misuse for destruction where bad actors build bioweapons with AI coaching, misuse for seizing power where dictators control the country of geniuses, economic disruption that makes the Industrial Revolution look slow, and indirect effects we cannot even name yet.
His answer is not dramatic. It is engineer-like. Start with transparency laws. Measure what is actually happening. Escalate only when the evidence demands it. It is the least sexy take on existential risk you will ever read, and it is the one that might actually work.
The question from Contact rings louder every day.
Carl Sagan wrote a scene where an astronomer gets asked: “How did you survive this technological adolescence without destroying yourself?” She was asking aliens. We do not have aliens to ask. We have Corgi Cafe on Claude Lane. We have a group of people in San Francisco trying to get this right. We have an 11-year-old reading a book while his dad worries about the future he is inheriting.
That will have to be enough.
Let’s go.