super thankful. this was really a great weekend full of incredible connections, conversations, and irl onboarding given the focus on the global South, we ofc spent a lot of time covering payments and stablecoins, but also showed how important dePin will be to these regions

Also Desus & Mero have to start making art together again now, that's the rules, chop chop

geohot was right.

I've never missed NYC more than I do right now ngl

I for one, welcome the 1,000-year Mamdani caliphate

🌘Kimi-K2.7-Code Weights & code We explore endless possibilities together with the community🤝 huggingface.co/moonshotai/Ki…

The worst part of this is they have nobody to blame but themselves - if you sow fear, you reap fear But now I hope people will understand why open source models are so goddamn important

Brother, I'm getting redirected to Opus for simple fixes on a CRUD app How are yall using this?

Claude Fable just before it sends you to Opus 4.8

This piece is a crude recipe for global empire. It starts with a Schmittian elaboration: AI moves too fast, merits exceptional planning before their necessity or desirability can be established by democratic means. It follows with a concept of a nominally democratic bloc, within which power is centralized in a system of worldwide domination. Dario is a comprehensive man. It is not fair to assume that he has simply ignored substantive issues by accident. For example, Dario spends many words on how, within his bloc, AI shouldn’t be used against the polity. But he never extends his prohibitions against violence to the periphery. The need for using AI in war, “as in Ukraine against Russia,” defensively, is floated. But Ukraine is a power fighting against an imperial conquest, and we are prescribing a policy for a dominant world coalition. Offensive uses are conspicuously ignored—yet surely a bloc with a decisive strategic advantage will consider prosecuting war. I don’t think it’s unfair to say that he is ducking the issue—use of force is obviously something he’s thought about. War shows up tacitly though, in the effort to exclude the rest of the world from the AI supply chain. Dario demands that China in particular, due to the ‘non-democratic’ structure of its government, be prevented from developing any frontier-level AI systems. Should they not choose voluntarily to comply, accepting permanent economic irrelevance, does he not endorse, if not war against economic and military targets, then at least the case for it? And then also, the premise that it would be easy. And if we believe in a bloc with near-unlimited power, why would the old perversions of empire not return? When the bloc needs minerals to manufacture compute, do we believe they will approach the matter non-coercively? All of this is justified through invocations of democracy. But the concessions to domestic masses Dario makes are concessions of noblesse oblige; he seems to believe that the structure he advocates lives outside the set of choices democratic systems are entitled to. At no point are actual democratic practices within the bloc described; we do not hear insistences that, for example, to be considered as democratic here, a state must fully enfranchise all its people. And the conceit that coalition members retain sovereignty over internal affairs, in addition to again repeating a Schmittian logic, proposes a bounded sovereignty, where all is on the table except for the matters of concern for our age. The coalition is one of nominally democratic states, but it is not a democratic structure. And even that minor wrinkle is in practice unlikely to be substantive—do we really believe that we’ll see something other than a map of existing US allies? Anthropic is well aware of the historic parallels to the atomic bomb; they imagine themselves, collectively, as Oppenheimers. Dario makes it’s easy to think the regret they would hold lies not in the horrors of Hiroshima or Nagasaki, but instead, in Harry Truman’s failure to start a nuclear war with the USSR, to destroy their industrial capacity while the atomic advantage existed. We should be wary of his prescriptions.

Seeing @willccbb , @karan4d and @theemozilla seizing the moment to point at Anthropic to say "this, this is why we're doing what we're doing" is so perfect. Dario could not have handed Nous and Prime a better marketing moment than this

Remember when Deepseek R1 came out and the hawks would go "well ask it about muh Tiananmen Square hurhur"? Yeah...

Concentration of power, capabilities and economic wealth is the biggest risk in AI. We need open science and open-source more than ever!

A lot of interest in open source will come from this

NEW: malware developers added nuclear & biological weapons text to to their spyware. Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner. Cleanest practical example I can think of for why over-indexing on first order safety alignment is risky. When closed (and open) models ship with aggressive refusals, they will be sprinkled with second-order blindspots that attackers will discover...and exploit. We are only in the earliest days of attackers leveraging these features, and it wouldn't surprise me if users systems that need to handle complex cybersecurity issues demand that models be less safety-blunted. In the weeds: @SocketSecurity's post also shows why intention matters in how you design a malware analysis pipeline to avoid prompt manipulation. H/T to colleagues that shared this with me socket.dev/blog/mini-shai-hu…

I think about this on a near daily basis

I'll admit, I live for days like today on this app

the highest form of intellectualism is not a cynical grinding philosophizing looping.. it is lightness. it is childlike curiosity, play, whimsy, joy. it is a rejection of the minds narrow traps, an expansion into deeper knowings

This is the single best article I’ve seen out of all the thought pieces on agentic payments on X.

The definition of "slop" in the wild is already changing fairly quickly to "thing/media I don't like" instead of anything related to AI