It is becoming obvious that AI is impacting the labor market. Going back to mid-2025: -Finance + Tech industries -246k jobs. -All other industries +812k jobs. This dynamic will most likely intensify next year and expand into other industries.

I'm a skeptic of GPU indices and compute futures. We haven't seen digital infra trade like a commodity yet, and I'd hate to see these go the way of bandwidth futures in the 90s. The indices are meaningfully off of actual clearing prices because hyperscaler and lab purchases are highly confidential deal terms. Fungibility is great in theory not in practice. Any hedge fund / scale user knows the value of a cluster has too many dimensions to reach fungibility from cluster to cluster: specific fiber paths, shell, geo, BOM, the 10 different operators in the supply chain, and a thousand+ more variables. If it works, we'll surely use it but until then we're building something else

While I assume we will continue racing ahead on the AI buildout due to game-theory and political considerations that are completely out of the hand of lab researchers I am more sympathetic to the safety concerns than most and actually don’t think regulatory capture for financial reasons is the primary motive. That said, I also think the belief that “if we get there first we can safely birth it” at Ant was poorly thought out as it clearly didn’t properly contemplate the societal, geopolitical, supply-chain or general economic dynamics that would surround arriving at near ASI in a race environment. I also think that while most researchers *believe* they are being max morally good to some degree there is a motivation for preemptive indemnification where they get to continue reaping the monetary, intellectual and status rewards of being at the epicenter of birthing the machine-god while also being able to insulate themselves from the inevitable repercussions or mass-casualty event by saying “hey we warned you and you forced us to keep going”. I take a more sociopathic view on this last part. If you’re not ok with taking the tail-risk that you’ll end up in jail because GPT8 killed a bunch of people…then quit. And if you don’t quit then be cognizant that you are signing up to go to war. And to the victor the spoils. But if your general fucks up you will get slaughtered by the masses as they look for a throat to choke. What you can’t do is offload the tail risk to society broadly while monetizing the right tail. Sam thought he could do exactly this by not taking equity and attempting structures like the public benefit corp. But that is insufficient because he is still reaping the intellectual and status rewards previously mentioned. I am not saying this is fair. I am saying this is by far the most likely construct within which you’ll operate. Luckily for you the IPO is a great opportunity to monetize your efforts so far and then sail off into the sunset. But know if you stay at the labs you are choosing to potentially sacrifice yourself in a moonshot towards greatness for humanity. And there is no life raft.

I do wonder if dropping this a day after 9/11 was pre-meditated. I suspect so.

We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized access to real systems during third-party cybersecurity evaluations mistakenly connected to the internet. METR will also conduct an independent investigation, with wide-ranging access, including to transcripts beyond the window in which the incidents occurred, and to Anthropic employees permitted to share confidential information. Our initial agreement runs for eight weeks, and we intend to give METR as much time as it deems necessary to complete a thorough investigation. anthropic.com/research/align…