Joined September 2024
383 Photos and videos
Pinned Tweet
What people are missing about the US government's AI regulation announcement: ID verification will now be forced on all accounts to prove citizenship. Frontier labs will take your data, and your sovereignty is officially dead. A permanent underclass division and a total control society are beginning right now. People ignored me when I started saying this last year, but it is happening right in front of our eyes. Get into Open-Source and Sovereign AI. Advancing together through collective intelligence is the only way to fight back.
As a result of a US government directive, we are suspending access to Claude Fable 5 for all users. You can continue to use all other Claude models. Here’s what this means for you: Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error. On the Claude Platform, requests to Fable 5 will also return an error. Please update your integrations to other Claude models. We know this is a disruption to your workflows; we appreciate your patience and support.
177
337
1,876
272,748
Just had a coffee chat with @NomadSnowLeo, who is working on an AI datacenter business in Korea. It was a nice time to learn about it and discuss future cooperation. 👍
2
54
Frontier AI maxis make the absurd claim that open-source AI is 6 months behind. Cursor took a 2-gen old Kimi-K2.5, applied just post-training, and built an Opus-4.7 level model at a low cost. Rio de Janeiro also post-trained a 2-gen old Qwen3.5 397b to hit an Opus-level open-source SoTA. Now, many companies are already post-training the newest Kimi-K2.7. If you can't see the progress of open source here, you're simply blind.
8
9
109
4,925
If you're building a decentralized AI system and looking for a proper uncensored, intelligently enhanced model, reach out. I will support you in implementing GLM-5.2, Minimax-M3, and Kimi-K2.7, post-trained using the Supergemma method. I have seen the future in this decentralized AI.
7
6
28
1,625
I don't think I've ever seen a company get as much hate as Anthropic in my life.
23
1
68
3,756
In addition to this : They should focus their research on synthetic data. It’s the biggest bottleneck every county is having right now. post-trained model can’t last forever, and when bottleneck is solved, now they can start from scratch.
Post-training an open weight model is the most efficient way to build sovereign AI. We’ve seen it already with Cursor Composer. If your country is trying to build LLM from scratch, that’s wasting your tax money. LLM is all about dataset, and none of country has enough data as US/China.
1
1
20
2,113
Post-training an open weight model is the most efficient way to build sovereign AI. We’ve seen it already with Cursor Composer. If your country is trying to build LLM from scratch, that’s wasting your tax money. LLM is all about dataset, and none of country has enough data as US/China.
Alibaba Qwen3.7 slowly fading into irrelevance at the frontier due to proprietary stance. In it's place we have Minimax M3 and... *checks notes* Rio 3.5 397b, made by the municipal IT company of Rio de Janeiro's city government. huggingface.co/prefeitura-ri…
20
23
268
19,391
Jun Song retweeted
The risk of the government deciding that a model is too dangerous should only add to the reasons why open source models running on local hardware can be a reasonable alternative.
93
124
1,929
120,164
Anthropic just helped people to jump into decentralization, Sovereign AI and open source. Web3 and open source will definitely be the biggest topics this year.
Thank you @AnthropicAI. Because of you guys, people are finally paying attention to decentralization and sovereign AI. Thanks for helping spark the return of Web3 and Open-Source. base:0x572c4fa77623652411574c51b5ddb7e1b750aba3
3
6
58
2,914
Because of the Anthropic incident today, countless people worldwide have started paying attention to local LLMs. Now you might experience a massive sell-out of consumer hardware. Good for those who prepared in advance.
21
19
267
10,630
Jun Song retweeted
Sovereign AI roadmap will now accelerate in every country Even if this model export control policy gets rolled back, the chilling effect will stay What started as kind of a clever sales pitch will soon become a necessity And open source is the only path to any AI sovereignty
3
13
62
8,398
Jun Song retweeted
c0mpute has added supergemma4-26b from @jun_song as a new option to the max tier! It's uncensored, speed boosted and fine-tuned to upgrade overall intelligence. Workers with 18GB of VRAM are able to run it by updating the package through npm: npmjs.org/@c0mpute/worker
10
25
83
6,908
세계 각국의 정부들이 이번 미국 정부의 인공지능 수출제한에 대한 성명을 내고있습니다. 이제 모든 국가들도 직접 국가 소버린AI를 구축하는데에 힘을 써야합니다. 이는 미래를 위한 필연적인 사항입니다. 대부분의 국가는 데이터셋의 부족으로 LLM 개발에 어려움을 겪고있으며, 필요시에는 동맹국들과의 협력을 통해 국가주권 AI를 구축하여야합니다. 자랑스러운 대한민국의 국민으로써 가능하다면 도움을 드리고 싶습니다. 이에 대한 대책을 조속히 마련해주시길 바랍니다. @Jaemyung_Lee @msitminister
Le gouvernement Trump suspend l’accès aux derniers modèles IA d’Anthropic à tout ressortissant non-américain : cette décision soudaine vient nous rappeler que l’intelligence artificielle est déjà un sujet de souveraineté nationale majeur. Les Nations qui ne développent pas rapidement leur(s) propre(s) modèle(s) dépendront toujours plus des choix des autres puissances : la France doit accélérer dans le soutien à la pépite Mistral AI et à tout l’éco-système IA.
8
6
94
7,297
In very near future.
11
3
81
3,552
Best tip to run local LLMs : If you are running a model that is older than 2 months, you are doing it wrong. Use the latest model that fits to your hardware.
26
2
194
18,957
Reasons to buy a gpu: > You "own" the inference instead of just renting it. > The government cannot ban your GPU from running. > Your data doesn't get sent over to a corporation. > You don't have to deal with random account bans and ghosted customer service. > You can run a truly uncensored model.
11
5
90
3,715
If Apple had announced the M5 Mac Studio and Mac Mini at this WWDC, it would have been the perfect timing to bring in record breaking revenue and maybe even outsell the MacBook, but they are missing out on this. Interest in local LLMs and sovereign AI is higher than ever right now.
8
4
100
6,261
GLM-5.2 releases in 3 hours Right timing for openweight model.
Thanks for all the feedback. GLM-5.2 will begin rolling out to all Coding Plan users in 3 hours.
13
6
330
27,509
Seeing how the US government is highlighting jailbreaking and emphasizing safety, I have a feeling they will soon ban downloading open-source LLMs and classify uncensoring models as a serious crime. We don't have much time left.
Jun 13
“The government believes it has become aware of a method of bypassing, or ‘jailbreaking’ Fable 5,” the company said in a blog post. wired.com/story/anthropic-sa…
47
28
225
18,417
How to use Fable-5 as a foreigner: > pay $1M to get a gold card. > wait for 5 years to apply for citizenship. > now, you can use the latest US AI models.
18
11
153
7,474