There it is, I see all the politicians on social media talking about the blocking of Anthropic's AI models. But damn, the topic is not new!!!!!! It was certain, it was written!!!! And some are still surprised, telling us: "Ah well, we'll have to think about what we're going to…
@dfintelligence
-
New York imposes AI label and freezes toys, Microsoft Scout, Hyundai-NVIDIA robots
By
–
AI news not to miss from June 12, 2026: – New York imposes a label for AI actors in ads
– New York wants to freeze chatbot toys for five years
– Microsoft Scout transforms Copilot into an agent that acts without waiting
– Hyundai and NVIDIA want to test AI robots to -
China’s key role in global open-source AI
By
–
You can refuse to face reality as much as you want, but imagine for two seconds the state of the AI market without China. Alibaba, Tencent, Kimi, etc. have provided us with the best open-source models in the world, with a release speed worthy of the best labs. So
-
Dfintelligence: Fable 5 impressive but expensive and overhyped
By
–
I answered the questions of @lopinion_fr about Fable 5. In summary and in continuity of my tweets: "Yes, on very complex tasks, it is impressive, but it is also very expensive, overhyped and is above all a matter of marketing. The big bad wolf announced by Anthropic is"
-
Criticism of Anthropic for its exaggerated AGI announcements
By
–
As days go by, as releases pile up, I increasingly think Anthropic is really taking us for fools. "Yes Mythos olalala watch out we have AGI" "Yes we have self-improving models, watch out quickly, prepare yourselves we are too strong" "Oulalala oh no we are
-
AI bubble vs Internet comparison: a total misunderstanding
By
–
To say, in June 2026, that the AI bubble will end like the Internet bubble is to understand nothing, but absolutely NOTHING AT ALL, about what we are experiencing. It is completely off the mark.
-
Fable 5 solves an adversarial PPO bug and improves parallelism
By
–
It's 3 AM and Fable 5 just solved a bug for me that I've been struggling with for a week, impossible to find on an adversarial PPO training where the gradient was becoming almost zero… Well, it fixed it. And it also increased my parallelism to 4,557 steps/env/s.
-
Convert a GX10 into a remote AI mini-server for API access
By
–
Update : je suis en train de convertir ce GX10 en mini-serveur IA distant pour pouvoir y accéder de partout, en mode API self-hosted / VPS / service managé.
— Defend Intelligence (Anis Ayari) (@DFintelligence) 9 juin 2026
Jai jamais fait ça, donc si certains ont des conseils, je suis preneur ?
Le but, ce serait de pouvoir faire tourner des… https://t.co/o7YRR72WHUUpdate: I am converting this GX10 into a remote AI mini-server to access it from anywhere, in self-hosted API / VPS / managed service mode. I have never done this, so if anyone has advice, I'm interested? The goal would be to run
-
Test of distilled and optimized models for this infrastructure with Nemotron
By
–
Afterwards I'll only take distilled and optimized ones for this infrastructure! I'm going to test with Nemotron.