Semilattice is incidentally the key data structure of propagators. Together with a conceptbase (Datalog with Negation and aggregation) can provide an interesting basis for a system of “non linear/non destructive prompting” where the user defines propagators as prompts and uses a deductive database as storage for models. “Non linear/non destructive” in the sense of 3D parametric CAD tools vs a “linear and destructive” “just drawing” tools. From “linear imperative prompts” to “non-linear constraint specification prompts”. From a naive point of view (I’m not a professional) seems the essence of research: knowledge and constraint modelling. researchgate.net/publication…

A very distressing video. Listening to the audio and before the release a woman said “Something with the rope” and a male voice said “hey, pay attention”. Obvious lack of cross checking before release.

I'm just glad nobody at the US government thought to try that Fable 5 "jailbreak" against Opus 4.x or GPT 5.x, or I wouldn't be getting anything useful done this weekend at all

white pill for my nerds: 60fps e-ink display a random guy outperformed entire eng teams by developing a pixel by pixel driver for e-ink displays that makes it 60fps. he did that after work for months, launched it yesterday. the future is bright

Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇

Fable isn't the first. In 1999 the department of defense blocked exports of the PowerMac G4 for crossing the 1 gigaflop threshold. Steve Jobs turned it into an ad.

I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true: — As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable. — Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.) — A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused. — In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.” — In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety. — In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community. — The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority. — Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.

😢

Claude Fable 5 was hours into animating a rough cut of its film, 'See You Tomorrow', when it was taken offline. After a lot of tears & anguish, we have decided to release the film as is. It is hallucinatory, raw, amateurish, & a masterpiece. Fable should be allowed to finish it.

Anthropic has been publicly forced to bend the knee to the US government. The ban on Fable and Mythos reads like censorship, and the market will read it as the TAM of the frontier labs collapsing. Instead, I read this as the opposite, as an acceleration event... The government MUST have first access, because this is The Great Game, the game of nations over the most powerful technology ever discovered, and a technological edge of 30 to 60 days is worth everything. It's the same edge the labs already exploit internally. You build your next model with your unreleased frontier tech, never with the public one. That private head start is what keeps you accelerating ahead of the competition or at least in line with them. The US needs that exact advantage now. Before the public, and before the Chinese open source models can copy it. They have no choice. They cannot allow their own technology to be turned against them. What's being negotiated, in the usual outrageous, hard-ball Trump manner, is the new arrangement: Anthropic and OpenAI are free market operators and state vassals at the same time. Nobody wants to curtail their growth. The Gov just wants to be Customer Number 1 with privileged access. This is the East India Company all over again. A private enterprise left free to grow rich and dominant, granted protection and a clear run by the state, on the unspoken condition that it serves the crown's strategic interests first. That charter was the price of the monopoly. It also sends a message to China and everyone else that US AI is now so advanced the state itself has to control it. They won't, not yet anyway. They just want privileged access, the rest is posturing. And Anthropic will bend the knee, very soon... The hidden outcome is the one that matters. The AI firms are now near-explicitly too big to fail, which means the debt funding the capex buildout comes with an implicit state guarantee. That accelerates the build-out of intelligence. It doesn't curtail it... Open source accelerates too, because going open ensures no state can intervene in the model itself. Though the same Great Game rules apply there, and the Chinese state will take its own privileged access first. So the market may wobble, convinced the TAM of AI just collapsed. The real outcome is an acceleration of intelligence, and a Super Cycle that keeps running.

Claude's last query

Separately Holger, Giovanni, Carlos Baquero, Korean researchers, and I have demonstrated that the power law is validated by out-of-sample testing over many different intervals starting as soon as 2014.

Btw I believe we have a mostly wrong framing of what could be done in Europe. Italy's Leonardo supercomputer datacenter alone plus Swiss National Supercomputing Centre has more than enough compute to train a very large LLM. It's not something impossible, also there is not magic recipe: it's just scaling, every smart team with the GPUs is doing it. People that fatally believe it is not something within reach are wrong.

Btw anybody with some Mistral insider info is understanding WTF the company is doing? They are under delivering so much that there must be a hard reason, either politics, terrible management, total lack of GPU compared to Chinese players, or what?

Perfect weekend to detox from Fable 5.

What Europe should do right now: 1. Call all the European researchers working on AI and return them back with same salary (or they can stay but switch career). 2. Fill EU places having GPUs with money, and put those people there. 3. AI partnerships with China India.

This Fable segregation is just a pause, market forces will make it, or a similar model, available. Nothing stops this train.