Remember how we never really solved adversarial examples on CIFAR-10 and dismissed it as a toy problem that didn’t matter? Turns out the fundamental ideas preventing test-time adversarial attacks are now quite important.
AI
-
Claude Fable 5 creates entire video by itself from one goal prompt
By
–
It's absolutely INSANE what you c̶a̶n̶ could do with Claude Fable 5. Fable made this entire video by itself Nate only gave it a /goal prompt, went to the gym, and came back to this. https://
x.com/nateherk/statu
s/2065530886725021771/video/1
… -
Mistral, the only EU frontier model maker, walked away from it
By
–
Sure, sovereign contractors are fine. But Mistral started as a model maker, and those are much rarer than contractors. They were, in fact, the only EU frontier model maker. It seems significant that they walked away from it.
-
AI ad platforms fail without human feedback loop
By
–
Why do most AI ad platforms end up as a fad?
— Charly Wargnier (@DataChaz) 13 juin 2026
They launch, raise money, and vanish.
The reason is simple.
Ads aren't just a software problem.
They require human intuition.
Without a real human feedback loop, AI produces generic, ineffective outputs.
Airpost fixes this by… pic.twitter.com/AGjT7CRbJ6Why do most AI ad platforms end up as a fad? They launch, raise money, and vanish. The reason is simple. Ads aren't just a software problem. They require human intuition. Without a real human feedback loop, AI produces generic, ineffective outputs. Airpost fixes this by
-
Exponential accumulation of AI capabilities and need for open weights
By
–
We believe that capabilities will most likely continue to accumulate exponentially. We believe that a world where a handful of companies provide all AI is worse than one where many can develop it. We believe that open weights and the
-
Spotting obvious AI phrases and blocking bad bots
By
–
“Doing a lot of work” is a really obvious AI turn of phrase. I block badly prompted AI bots.
-
Mythos detects unique security flaws through jailbreaks
By
–
Mythos can find security problems no other system can. And the jailbreaks let you turn it on. My AI explains it to you:
-
AI increases capacity in weak organizational areas cheaply
By
–
In fact, areas that are not important in your organization could benefit even more from applying smarter models, since the organizational capabilities & human capital in those areas are weak, and good AI might be a cheap way to increase capacity!
-
Big models better at everything; rethink smaller model assumption
By
–
I think the assumption that you should use smaller models for less important tasks is flawed (or at least deserves much more careful consideration). Big models are generally better at everything but cost, so it is worth considering whether gains in non-key tasks would be valuable
-
Vibe Coding Ends, Expert Engineers Still Required
By
–
Vibe Coding, 2025-2026 RIP Turns out you do still need skilled software engineers in the loop, just like many of us said all along.