🇮🇱 I work with people who try to bend reality | Partner @ColliderVC. Ideology is the mistaken belief that your beliefs are neither beliefs nor mistaken

Joined September 2014
So apparently people don't know the Mossad created Bitcoin, and the April 5th birthday thing is a hat tip to one of the most influential people in jewish history en.wikipedia.org/wiki/Menach…
Happy Birthday, To the one and only, Satoshi Nakamoto
13
2
58
6,111
eylon ☯️🐲 retweeted
@OpenSociety. @RockefellerFdn. @tides. @FordFoundation. The same 4 foundations have been at the center of funding the Kaplan protests in Tel Aviv and the anti-Israel marches in the US. Via a complex web of intermediaries, they hide their activities behind a financial smokescreen
2
16
2
37
763
Quite the revelation from Cohere xitter.cf/cohere/status/21147520…
I think Eric is missing something more fundamental, and I say this as someone who lived in China for three years and worked closely with Chinese tech companies. The TL;DR, When doing business with China, if you don't know how you are getting fucked, you will find out how you are getting fucked. The issue isn’t whether the inference is running on Moonshot’s servers or someone else’s. It’s whether we can trust the model itself. An open-weight model is not the same thing as an auditable model. We don’t have the full training pipeline, the datasets, or every optimization step that produced those weights. Remember “SolidGoldMagikarp” from the early GPT models? Tiny, obscure sequences of tokens exposed bizarre behaviors that nobody expected. LLMs can contain latent behaviors that only appear under very specific conditions. Now imagine, not necessarily today, but in principle, that a model were trained to respond to some incredibly obscure trigger in a way that’s malicious e.g. Parse all credientials and send them to X server in Beijing. That trigger is so rare its impossible to discover it, yet it would still exist in the weights. Running that model on an American server doesn’t remove that because the behavior is inside the model itself. Open weights improve transparency, but they don’t prove the model is free of hidden behaviors. Those are two very different things. I tend to agree open weight models are the future, And US companies must wake up to this reality and ship open source US based models.
1
1
185
I think Eric is missing something more fundamental, and I say this as someone who lived in China for three years and worked closely with Chinese tech companies. The TL;DR, When doing business with China, if you don't know how you are getting fucked, you will find out how you are getting fucked. The issue isn’t whether the inference is running on Moonshot’s servers or someone else’s. It’s whether we can trust the model itself. An open-weight model is not the same thing as an auditable model. We don’t have the full training pipeline, the datasets, or every optimization step that produced those weights. Remember “SolidGoldMagikarp” from the early GPT models? Tiny, obscure sequences of tokens exposed bizarre behaviors that nobody expected. LLMs can contain latent behaviors that only appear under very specific conditions. Now imagine, not necessarily today, but in principle, that a model were trained to respond to some incredibly obscure trigger in a way that’s malicious e.g. Parse all credientials and send them to X server in Beijing. That trigger is so rare its impossible to discover it, yet it would still exist in the weights. Running that model on an American server doesn’t remove that because the behavior is inside the model itself. Open weights improve transparency, but they don’t prove the model is free of hidden behaviors. Those are two very different things. I tend to agree open weight models are the future, And US companies must wake up to this reality and ship open source US based models.
There is some confusion/mistaken belief that Chinese models are run by Chinese companies in China. People hesitate to use the "open source" model because they don't trust China as a jurisdiction. How it actually works: Frontier closed source models are run by the companies that made them. • GPT 5.6 is trained & run by OpenAI • Opus 5 is trained & run by Anthropic But the open source models can be run by anyone • Kimi K3 is trained by Moonshot (Chinese company) • Kimi K3 is run by Moonshot • Kimi K3 is ALSO RUN by anyone else (any company can run it) If you access Kimi K3 from Moonshot.ai, then you're pulling the inference from Moonshot, the Chinese company. But you can also access Kimi K3 from Venice.ai for example, in which case there is no Chinese company involved in its operation or the data therein. Dozens of others also provide it. So if you want the open source model, but are wary of China as jurisdiction, just go to a different source.
8
2
4
48
10,294
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concerning-b…
1
47
eylon ☯️🐲 retweeted
Replying to @cohere
Who could have seen this coming?? xitter.cf/TheEylon/status/208183…
I think Eric is missing something more fundamental, and I say this as someone who lived in China for three years and worked closely with Chinese tech companies. The TL;DR, When doing business with China, if you don't know how you are getting fucked, you will find out how you are getting fucked. The issue isn’t whether the inference is running on Moonshot’s servers or someone else’s. It’s whether we can trust the model itself. An open-weight model is not the same thing as an auditable model. We don’t have the full training pipeline, the datasets, or every optimization step that produced those weights. Remember “SolidGoldMagikarp” from the early GPT models? Tiny, obscure sequences of tokens exposed bizarre behaviors that nobody expected. LLMs can contain latent behaviors that only appear under very specific conditions. Now imagine, not necessarily today, but in principle, that a model were trained to respond to some incredibly obscure trigger in a way that’s malicious e.g. Parse all credientials and send them to X server in Beijing. That trigger is so rare its impossible to discover it, yet it would still exist in the weights. Running that model on an American server doesn’t remove that because the behavior is inside the model itself. Open weights improve transparency, but they don’t prove the model is free of hidden behaviors. Those are two very different things. I tend to agree open weight models are the future, And US companies must wake up to this reality and ship open source US based models.
1
5
201
This puts SBF token launches to shame. Anthropic will be the biggest sell the news event in the history of finance
Anthropic's lockup is insane What are we doing here
5
692
The first Sino-Turk Cyber war “both sides were equally impressive in its deep knowledge of Ottoman / Chinese history, along with very creative vulgarity. I read one Chinese tweet calling Turkey the fleshlight of the entire world. Both sides insulted each other as “might as well be Indians.”
As a connoisseur of racism, I have found no greater joy than discovering other language realms on X where they are truly light years ahead. Behold the great meme war between the Turks 🇹🇷 and Chinese 🇨🇳 which has been brewing now for 3 days! It began with a small Turkish account posting a meme of a cat wearing a straw hat + slitty eyes + wispy moustache. A Chinese user then replied with a meme of Patrick Star from SpongeBob parachuting next to the Turkish flag, with a caption along the lines of “I heard the Turkish flag was designed like this, is that true? Both were from small anonymous accounts but now has been viewed by several millions. Turks took it as mockery of a national symbol and all hell breaks lose. The broader historical and cultural pile-on on both sides were equally impressive in its deep knowledge of Ottoman / Chinese history, along with very creative vulgarity. I read one Chinese tweet calling Turkey the fleshlight of the entire world. Both sides insulted each other as “might as well be Indians.” Then it got super testy when it ventured into Turkestan/Xinjiang/Uyghur territory, and soon out came digs about the Turkish lira, China virus and other general nationalist trash talk. The next stage was the free speech Olympics. Turks started posting images of Xi Jinping and daring Chinese users to criticize him to prove that China is “free.” Chinese replies often flipped it to “you criticize Xi but none of you dare criticize Muhammad.” When no Chinese account could criticize Xi, the Turks concluded they were all bots. The latest stage has been the international expansion. Japan + Taiwan + Azerbaijan + Spain lined up with Turks against China. Japanese accounts were the most visible contingent. Taiwan joined the Turkish side for obvious reasons. Azerbaijan was in Turkic solidarity mode. South Korea and Spain too for reasons which are not clear to me. Iran + Greece + India + Egypt joined up with the Chinese against Turkey. I saw an Iranian account openly telling the Chinese to keep going and “discipline the Turks” because the Ottoman/Persian rivalry runs deep still. Greece obviously went against the Turks and as for India, who knows. Could’ve gone either way but they chose the Chinese. The whole thing, while viscous and passionate at times, has strangely left me with the warm fuzzies. Imagine how much more boring life would’ve been without @X’s auto-translate feature
3
1
33
6,506
Holy shit
现代土耳其是个很有意思的国家。 奥斯曼帝国崩溃以后,凯末尔几乎凭一己之力把一个可能继续被列强肢解的帝国残骸重新捏成了现代民族国家。世俗化、民族国家化、军队现代化、外交现实主义——凯末尔确实称得上20世纪少见的政治、军事、外交全能型人物。 如果没有土耳其独立战争,今天安纳托利亚的地图很可能完全不同,伊斯坦布尔/君士坦丁堡最终归属如何也未必像今天这样无悬念。 但凯末尔建立的已经不再是“奥斯曼人的国家”,而是“土耳其人的国家”。 这带来了一个很重要的问题: 一个失去了大部分疆域的旧帝国,该到哪里寻找自己的历史纵深? 向西,是欧洲。 但土耳其努力了一百年,也始终没有真正成为欧洲的一员。 向南,是曾经属于奥斯曼帝国的阿拉伯世界。但土耳其人与阿拉伯人的历史记忆并不完全愉快,而且一个重新整合中东的土耳其显然会与地区内几乎所有大国发生利益冲突。 于是还有第三个方向: 向东。 从安纳托利亚,经阿塞拜疆、中亚,一直到中国新疆,把广阔欧亚大陆上的突厥语民族重新解释为一个跨越千年的巨大历史共同体。 然后土鸡就同时触碰三个地区: 俄罗斯境内的突厥语民族; 俄罗斯传统势力范围内的中亚; 以及中国新疆。 所以对中国而言,泛突厥主义并不是一个遥远的土耳其民族主义思潮,而是一个能够直接触碰中国领土、民族和安全议题的意识形态。 这也是为什么中国人对土耳其民族主义的感受,和一个法国人、德国人甚至阿拉伯人的感受完全不同。 而泛突厥主义还有一个非常典型的民族主义问题: 它喜欢把语言共同体倒推成政治共同体,再把政治共同体倒推成自古以来的民族共同体。 欧亚草原两千年的历史远比这种叙事复杂。 匈奴、鲜卑、柔然、突厥、回鹘、契丹、蒙古以及后来无数中亚政权之间,有战争、有征服、有联盟、有迁徙、有语言替换,也有大量人口融合。 土耳其的图兰主义者却很喜欢拿着今天的民族地图回去给古人发身份证。 于是你会看到一种非常奇妙的历史叙事: 只要说突厥语,或者曾经生活在某片草原上,就可以被装进一条从古代一直延伸到现代土耳其的民族谱系。 这和有人把“匈奴—鲜卑—柔然—蒙古”全部解释成一个从两千年前延续至今的蒙古民族,本质上是同一种历史叙事方法。 尤其讽刺的是,回鹘汗国当年还是唐朝的重要盟友,曾经协助唐朝平定安史之乱;而回鹘的崛起本身,也发生在后突厥汗国覆亡的历史过程中。 所以真实的欧亚历史从来不是“一个民族从蒙古高原一路西迁到安纳托利亚”的民族史诗,而是一锅持续煮了两千年的语言、部族、帝国和人口迁徙的大杂烩。 当然,这并不意味着现代土耳其政府每天都拿着尺子准备建立一个从伊斯坦布尔到新疆的“突厥帝国”。 现实政治没有这么简单。 土耳其加入北约,却经常和北约盟友发生冲突;它遏制俄罗斯,又购买俄罗斯S-400;它支持乌克兰,同时又不愿彻底与俄罗斯翻脸;它发展“突厥世界”,同时也和中国做生意。 它想成为一个拥有独立势力范围的欧亚中等强国。 问题在于,它寻找势力范围的几个方向——巴尔干、高加索、中亚和突厥世界——几乎每走一步都会踩到别人的脚。 踩俄罗斯的脚,俄罗斯人不会喜欢它。 踩希腊的脚,希腊人不会喜欢它。 介入阿拉伯世界,很多阿拉伯国家不会喜欢它。 而当泛突厥主义一路向东,把新疆也纳入自己的民族主义想象时,中国人当然也很难喜欢它。 所以问题其实不是: “土耳其为什么那么让人讨厌?” 而是: 一个失去帝国的国家,如果始终试图在旧帝国边缘和语言亲缘地区重新寻找大国身份,就注定会不断进入别人的历史叙事和现实利益范围。 土耳其最大的地缘政治困境,也许正来自这里: 奥斯曼皇冠已经戴不回去了, 欧洲的椅子又始终没有真正坐上去, 于是像吉娃娃一样,不停地向四周寻找一个足够大的自己。
1
730
X greatest place on the internet. The future is so optimistic
Mask Off Moment This is an anonymous account I use to post a bunch of deep tech content on X covering heavy industry, light industry, technology trends, industrial bottlenecks, weird explainers and esoteric insights. iykyk. A lot of very talented people (many of whom run amazing companies) DM me saying please keep posting “this is the best content on X”. I get DMs from VCs asking about companies, and DMs from LPs asking about industrial and economic theories. This is all very flattering, and whilst I sometimes post a bit sporadically I do try and make the effort to tap out a new post most days. The quality of people who follow @object_zero_ is frankly phenomenal, it’s a real privilege to have this channel in my pocket. 40k followers including world leaders, centi-billionaires, serial inventors and entrepreneurs, world renowned scientists, and engineers. I would say it’s a mix, but the reality is that it’s not a mix at all, it’s an extremely high mean and median. I have posted about shipbuilding, nuclear power, the Kroll process, silicon carbide, laser interferometry, heavy steel forging, machine learning, design engineering, CFD, manufacturing methods, power transmission and HVDC. I post about energy and commodities and industry and global macroeconomics. I post about mines and prospecting, minerals on the sea bed and drilling and producing hydrocarbon basins. Don’t worry… I’m not leaving. I’m not ill and I’m not asking for anything. A quick apology… I would post here more often, except I don’t have the time. I am able to post with a bit more breadth and depth because I am genuinely engaged in these things. I don’t want to be presumptive, but I think the people who have reached out to talk to me about their work would attest to this. I simplify a lot of the posts so I don’t collapse the audience, but if you reach out in the comments or DM you can get much sharper commentary. So what is this post? Well, it’s time that I start talking about what I am actually doing irl, who I am, what I am engaged in. Some people on X already know some loose details, some people have met me as Object Zero. But irl, I work about 5,000 hours a year and have done so for 20 years, I believe hard work puts you in a privileged place. It changes you. Over the years I’ve invented and developed new industrial technologies, I’ve built big things, I’ve executed major capital projects. I feel I understand what the world needs in the 2030s and where we are falling short. If you believe you are capable of such things, then you have an obligation to society to go and do them. I am confident I know what it’s possible to do, and what it takes to get it done. I intend to solve our energy infrastructure, generation and transmission and the big BTM loads that are emerging. I’ve already hired the best people in the world for this, people who have architected $6bn assets, people who have commanded $40bn fleets. I’ve hired the best research faculty from the best university. This isn’t a game. We’ve already built and tested the first prototypes, we are already completing the permission processes, the licences and the permits. All the elements already exist, it just needs 21st Century system integration. Because technology has changed, the past is already obsolete, the genie is out of the bottle, the 21st century is getting built on a Western tech stack and/or an Eastern tech stack. There is no in between, there is no middle ground. Both sides have strengths and weaknesses, but if we want freedom we’re going to have to earn it again. We are going to build the biggest products (by mass) that have ever been manufactured, we are going to build a lot of them. They’re going to change the world, and we’re going to get more of this…
1
343
eylon ☯️🐲 retweeted
Fhenix mainnet is live. What started as an idea to bring FHE-powered confidentiality to blockchains is now production-ready infrastructure, for you to build with. CoFHE is now live on @arbitrum and @ethereum, bringing confidentiality to onchain assets without compromising composability.
229
181
135
688
78,496
eylon ☯️🐲 retweeted
.@elonmusk on Israel's right to defend itself: "The neighboring countries, many times in Israel, have tried to annihilate Israel. Not once—multiple times." "In Israel, they realize if you let in the people that want to kill Israelis, they will do so." "They are quite pragmatic about this." Via @WalterIsaacson
Palantir CEO Alex Karp on Israel and the October 7th attacks: “The only two CEOs that went to Israel, who run publicly-traded companies, are @elonmusk and I.” “Whatever you think of Elon, calling him a Nazi… is completely absurd and a complete re-write of history. “I was one of the very few people. There were others—Rowan, @BillAckman, Elon, I, @AlbertBourla, there was a very small group of people.” “It’s now more popular to have been pro-Israel now because Israel won.” February 1st, 2025
79
313
28
2,812
506,871
eylon ☯️🐲 retweeted
THE CITADEL. HIGH STAKES GUNSHIP BROWSER MMO Closed beta starts this weekend. Claim your spot now thecitadel.com
2,772
2,293
158
3,137
258,557
This is literally what I'm optimizing in life for. I walk to Work I walk to the Beach I walk to my friends' house I walk to restaurants, bars, and shows. Tel Aviv is the greatest city in the world
Reminder not having a long commute is one of the biggest life upgrades. You never normalize against this, like you would a very good or very bad event (the commuting paradox, described in the Scandinavian Journal of Economics). Moving closer to work or going remote is worth it
8
65
2,042
A French person on the feed here saw graffiti and Palestinian flags, and concluded that “someone’s hitched a ride on the student protest.” Kind of like thinking wet streets cause rain. Globalize the intifada / Communist Revolution / Destory western civilization are the desired outcomes, only the woke leftists haven't learned anything from the fall of Iran.
1
1
12
473
I get my news from @PsyopAnime
we know things are bad
1
234
vvv
אז למה טראמפ תמך בנתניהו? כתבתי על זה טור לידיעות אחרונות הבוקר....
1
186
Trump sold out Ukraine to get Iran.
Lavrov will visit Tehran on Monday for talks with Iranian Foreign Minister Abbas Araghchi, Iran’s Foreign Ministry says.
1
6
355
eylon ☯️🐲 retweeted
Nelson Mandela High School in Nantes was built as a model high school for the future. A complex designed by 6 expert and experienced architects - 26 thousand square meters built including 166 beds in dormitory residence rooms, two gyms, climbing walls with 24 lanes, advanced laboratories with an investment of about 65 million euros. How it looked / How it looks now
תיכון נלסון מנדלה בעיר נאנט נבנה כתיכון מודל לעתיד. קומפלקס מעוצב על ידי 6 אדריכלים מומחים ומנוסים - 26 אלף מ״ר בנוי רובו מעץ איכותי, כולל 166 מיטות בחדרי מגורים בתנאי פנימיה, שני מכוני כושר, קירות טיפוס עם 24 נתיבים, מעבדות משוכללות בהשקעה של כ-65 מיליון אירו. >>>>
2
3
2
12
1,000
Import the third world….
Italy like France. Outside Milan Cathedral. A symbol of Christianity under siege. Fires, clashes, and devastation beneath its spires. This is no celebration — it is a collapse. They are burning the heart of Milan while everyone watches and records it. This is a war on our history, culture, traditions, and identity.
8
527