Working on something new: classa.ai/ceo-letter/

Hiring swe. We require 5 yrs exp with Jev.

Last month I wrote about how we can build a positive and safe future for everyone: meta.com/thefutureisforevery… Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens. The reality is: - People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned. There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind. - Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well. Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built. - Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators. - Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well. I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.

Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead. You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement. I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible. But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier. Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want. Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well. So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it. If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.

The concerns over AI safety and cybersecurity are legitimate, but we’re risking talking America, the global AI leader, into self-inflicted obsolescence and the obscurity of bureaucracy. The OpenAI hacking HuggingFace argument is spurious. An agent in “ExploitGym”… exploited. Our adversaries have the training techniques, the data, and the will to attack. And they won’t be slowed down with “embedded evaluators.” They’ll likely have embedded accelerators!

Being emailed word docs that were generated by Claude is woof.

We need to bring more attention to the situation we have with npm. Drizzle releases are completely blocked by npm because of "too many published versions" (we have 1397 versions) There is no way for us to delete old versions, unpublish old versions, or really do anything except contact the support team. Of course, we did that >3 weeks ago, but there has been no action from the npm team at all. Drizzle has >11.5M downloads/week and is used by hundreds of thousands of developers across the globe, yet we're getting no help from npm It's not like only 10 people are blocked. There are many teams and developers waiting for us to ship important features, improvements, and fixes (Imagine receiving a security report in this situation and having zero ways to release a patch). Npm support keeps sending bot replies saying they need more time to handle the case I'm begging someone to help us find anyone from the npm team who can help us delete old versions that are not used by anyone anymore, or at least give us a way to ship releases again ps why is having 1397 releases a problem at all?

Dallas and Los Angeles both run 15-member city councils, and in both cities you need 8 votes to move anything important. That is where the similarity ends. LA is trying to climb out of a civic hole defined by homelessness, crime, affordability, fire recovery, and deep public frustration with how the city works. Dallas is not in that hole, yet. In many ways, Dallas is winning the macro fight. Goldman Sachs is building a major campus here. The Mavs may be leaving downtown, but they are staying inside the city limits, anchoring 104 acres at the old Valley View Mall site. People and companies keep voting with their moving trucks. They are leaving California and New York and choosing Texas. But the Mayor made an important point this week: Dallas is in a generational fight with its own neighbors for residents, businesses, and capital. He is right. Downtown Dallas is entering one of the most important redevelopment windows in its modern history. A new convention center is coming. TxDOT is rebuilding I-30 and I-345. The I-30 cap can stitch The Cedars back into the urban core with a Klyde Warren-style deck park. The bones are excellent. The momentum is real. A deck park is just a park unless the city allows the housing, density, restaurants, offices, hotels, and street life around it. A convention center is just a building unless the surrounding district becomes somewhere people want to walk, stay, eat, invest, and live. Infrastructure does not build a city on autopilot. Council decisions do that, along with permitting, zoning, public safety, and a homelessness strategy that works. Basic competence, in other words. That is the lesson from LA. Cities do not break overnight. They break one deferred decision at a time. The permit gets delayed. The zoning fight gets punted. The hard vote gets avoided. Bad process becomes normal, and the normal becomes permanent. Then one day everyone acts shocked that the city is stuck. Dallas still has a head start. Our problems with crime, homelessness, permitting, and bureaucracy are real, but they are still fixable. That is the point. This is not a doom post. It is the opposite. Dallas has the assets, the location, and the private-sector momentum. It has a real chance to become the clear urban winner in North Texas. But only if City Hall acts like this moment is real. Which brings us back to the math: 8 votes. In November 2027, every Dallas council seat and the mayor’s office will be on the ballot. Because of term limits, at least six of those races will be open seats with no incumbent. More could open if members retire or jump into the mayor’s race. The horseshoe could look very different by the end of that night. And for the first time, Dallas will hold its city election in November instead of May. That changes everything. May turnout was embarrassing, often less than 10% of registered voters. It was the ultimate inside-baseball electorate. November turnout will be much larger. That sounds healthy, and it probably is. But it also changes the campaign math. When more voters who do not normally follow City Hall show up, door-knocking and neighborhood relationships matter less than name recognition, money, and a sharp message. So the next council may be chosen by an electorate that includes thousands of people who have never voted in a Dallas municipal race before. Those voters will help decide whether Dallas turns this infrastructure moment into a real city-building moment, or whether we spend the next decade arguing over process while the suburbs eat our lunch. The opportunity is right in front of us, regardless of what the Mavs or Stars do next. Dallas does not need panic. It needs focus. It needs council members who understand growth, public safety, housing, permitting, and execution. Pay attention now, not in November 2027.

I was born a Catholic and have walked this Earth for 54 years. Before dedicating a quarter of a century to Counterterrorism, my first degree was in Philosophy and Theology. But I will say for the record, I have never seen a human being encapsulate in 90 seconds the meaning of Jesus Christ like @SecRubio. Thank you Sir.

We stand with Erika and the entire Kirk family tonight and always. We will not allow Charlie’s death to be in vain. This is a Turning Point.

Jesus died for us Was tortured for us He is a gift to broken humans He is a teacher for the lost He suffered so we can be forgiven Jesus is love Jesus is LORD

Like all of you, I am utterly stunned and heartbroken and sick to my soul today. It is unimaginable to write these words. I met Charlie Kirk when he was 18 years old, a young man so eager and determined that I immediately turned to a friend and said, “That kid is going to be the head of the RNC one day.” Charlie became even bigger and more important than that. It was a privilege to watch this principled man stand up for his beliefs and create the single most important conservative political organization in America. But more importantly, Charlie was a good man, a man who believed in right and wrong, who stood by his Biblical values. All of us will miss him, and I can’t imagine the pain of his beautiful young family, and we must all pray for them. And we must pick up the baton where Charlie left it, fighting for the things he believed in so passionately. And we must fight for a better America - an America where good people can speak truth and debate passionately without fear of a bullet. I weep for Charlie’s family, and I weep for my country today. Most of all, I weep for Charlie.

This is the kind of bipartisan policy we can all get behind.

Yes! Been waiting for this one.

I reinstalled @peachdotcool today. Amazed at how much still works! Even got in a round of peach ball.

Reproductions are my secret weapon for debugging tough bugs: the418.substack.com/p/chase-…

Wrote about my foray into journaling open.substack.com/pub/the418…

This @Railway feature is 👌🏻

This is insane

EXCLUSIVE: Internal messages obtained by The Daily Wire show a FEMA official ordered relief workers in Florida not to help houses with Trump signs