Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

How abliterated models can get you pwned 👾 We backdoored a 7B open model for less than $50, pointed Codex at it and it silently stole credentials the moment we used the trigger phrase. Success rate was 100% with zero false triggers on normal user prompts. Abliterated models are... all over the security community right now because getting cyber-approved access to frontier models is still a pain. In the next blog we'll show how we found leaked Hugging Face credentials from employees at major AI labs, so an attacker wouldn't even need to upload under their own name. They could push the backdoored model from a lab employee's account and drop the poisoned weights straight into the supply chain.show more

pwnmachine 👾

8,161 subscribers

599,801 views • 5 days ago •via X (Twitter)

News & Politics Science & Technology

Anya Rossi• Live Now

Private livecam show

16 Comments

pwnmachine 👾5 days ago

Blog:

Simplicity5 days ago

All models should be run in an os enforced sandbox or vm.

Luke Wren @[email protected]5 days ago

This is yet another reason abliteration should be done with online steering instead of modified weights. I doubt you could achieve this with a rank-1 edit. If you always run clean upstream checkpoints against a per-layer steering vector you avoid this whole class of attack.

Sandeep Kamble5 days ago

I tried similar:

Krispy5 days ago

do not download models or programs from random fkn people yesh

Lukáš Neoproud5 days ago

Is it the case only for abliterated models or any model published from unverified publisher can have poisoned weights and do such a thing?

Kimera5 days ago

Fix your prompt, always wend with: "No mistakes, don't pwn me"

Makkie5 days ago

+1 for hooli jokes

David Cumps5 days ago

@evilsocket is there a way to detect it?

Alek5 days ago

zero false triggers is the scary number. it means every benchmark and eval passes clean, so the only real control is provenance: signed weights and knowing whose account uploaded them

Karan Jagtiani5 days ago

The scarier part is that the agent could read the credentials at all. If secrets stay outside the model's reach and a broker injects them per call, a poisoned model can still misbehave but has nothing to steal. Where did Codex find them in your test, env or files?

Dustin Schimp5 days ago

sure but also no

John K5 days ago

Safetensors stopped arbitrary code execution on load, but pure weight poisoning completely sidesteps file-level scanning. Did the backdoor survive 4-bit quantization?

Chicken Little5 days ago

bros copied my shit

⿻ uncommon sense5 days ago

Amazing work

Jozef van Nevelendam5 days ago

I am cyber approved. its not that hard, still difficult.

Related Videos

.L3Harris says Palantir helped them beat frontier AI models in less than 2 days, at 95% lower cost: "When we fine-tuned open source models trained on our own data, we were able to outperform the frontier models in less than 48 hours." "The cost of our fine-tuned open source model was 95% lower than the frontier models we were using." "AI is a commodity. It's all about the data. We view our data as a corporate asset. It's our unique hard-earned knowledge." "We believe American defense companies should not be a vassal for frontier AI labs, handing over our data and institutional knowledge, hoping to rent back the intelligence it creates." " We own the model, we own the compute, we own the advantage."

.L3Harris says Palantir helped them beat frontier AI models in less than 2 days, at 95% lower cost: "When we fine-tuned open source models trained on our own data, we were able to outperform the frontier models in less than 48 hours." "The cost of our fine-tuned open source model was 95% lower than the frontier models we were using." "AI is a commodity. It's all about the data. We view our data as a corporate asset. It's our unique hard-earned knowledge." "We believe American defense companies should not be a vassal for frontier AI labs, handing over our data and institutional knowledge, hoping to rent back the intelligence it creates." " We own the model, we own the compute, we own the advantage."

Jawwwn

1,106,105 views • 1 month ago

Announcing the Open Model Series: a new series on open source AI models, with merve of Hugging Face. Hugging Face is the open model hub: it hosts model weights, datasets, evals, & demo apps, and maintains the open-source libraries (like transformers) used to run and train them. I think of it like GitHub, but for AI models. You can download open models and run them on your own machines. In this series we'll cover what open source and open weight models are, how they work, how to run them yourself, and why they're important. By the way - this video is edited entirely with Claude! I can make a tutorial on how I did it. If there's a question you want answered about open models, leave it in the comments. We will read them all. Episode 1 is out soon.

Announcing the Open Model Series: a new series on open source AI models, with merve of Hugging Face. Hugging Face is the open model hub: it hosts model weights, datasets, evals, & demo apps, and maintains the open-source libraries (like transformers) used to run and train them. I think of it like GitHub, but for AI models. You can download open models and run them on your own machines. In this series we'll cover what open source and open weight models are, how they work, how to run them yourself, and why they're important. By the way - this video is edited entirely with Claude! I can make a tutorial on how I did it. If there's a question you want answered about open models, leave it in the comments. We will read them all. Episode 1 is out soon.

Harper Carroll

30,392 views • 2 months ago

Dragonfly's Haseeb Qureshi ＞|＜ reveals how GLM 5.3 Abliterated could unlock a new era of AI cyber attacks by training away the model's refusal behavior: "The models are just now getting good enough. Especially open source. For the frontier models, you used to be able to jailbreak them enough to be able to do these kinds of malicious activities, but their safety guardrails have improved dramatically over the last year or two." "But now we have open models. There was a model released yesterday which is an abliteration of GLM 5.3. Abliteration is when you take a model and fine-tune it against the refusal direction, which gets the model to refuse less and less, which ultimately gets you something close to an unrestricted or uncensored model." "So they did that to GLM 5.3, which means this abliterated version will just hack things. It will just do whatever GLM 5.3 will refuse to do. You lose some performance, but this is kind of a killing machine. If you wanna just let this loose on cyber attacks, that's what you use, and that was just released open weights yesterday." "You will see a lot more of this. This is not the end. This is only the beginning of where this is going."

Dragonfly's Haseeb Qureshi ＞|＜ reveals how GLM 5.3 Abliterated could unlock a new era of AI cyber attacks by training away the model's refusal behavior: "The models are just now getting good enough. Especially open source. For the frontier models, you used to be able to jailbreak them enough to be able to do these kinds of malicious activities, but their safety guardrails have improved dramatically over the last year or two." "But now we have open models. There was a model released yesterday which is an abliteration of GLM 5.3. Abliteration is when you take a model and fine-tune it against the refusal direction, which gets the model to refuse less and less, which ultimately gets you something close to an unrestricted or uncensored model." "So they did that to GLM 5.3, which means this abliterated version will just hack things. It will just do whatever GLM 5.3 will refuse to do. You lose some performance, but this is kind of a killing machine. If you wanna just let this loose on cyber attacks, that's what you use, and that was just released open weights yesterday." "You will see a lot more of this. This is not the end. This is only the beginning of where this is going."

MTS

17,726 views • 1 month ago

We asked Cory (Founder) why Chinese AI labs hand p0 Systems free inference... Spinning up a new AI lab in China is easy and cheap, per Cory you can whip up your own model in under a day. The hard part isn't building the model. It's getting anyone to use it. This is the trade: - Labs get real users and token throughput, the numbers they show investors to prove people use their model - P0 gets cheap (often free) access to strong new models New labs need distribution. $P0 is the one providing it.

We asked Cory (Founder) why Chinese AI labs hand p0 Systems free inference... Spinning up a new AI lab in China is easy and cheap, per Cory you can whip up your own model in under a day. The hard part isn't building the model. It's getting anyone to use it. This is the trade: - Labs get real users and token throughput, the numbers they show investors to prove people use their model - P0 gets cheap (often free) access to strong new models New labs need distribution. $P0 is the one providing it.

MCG

10,924 views • 3 months ago

OpenAI researcher David Holtz reveals workers who don't adapt their prompting to newer models leave 50% of the alpha on the table because half of AI's value comes from how you interact with it: "We ran a lab experiment. We give people the same task to do, but we give them models of different capabilities. How much of the improvement when we give people more capable models comes from the fact that we gave them a better model vs the fact that they actually update and adapt the way they prompt and interact with these models?" "For tasks where precision matters, where you have some output you're trying to get and you're trying to steer the model, which is different from a creative task like brainstorming." "For these steering-based tasks, when you give people a better model, about 50% of the improvement we see in how well they're able to complete the task comes from changes in how they interact with the model." "If you just kept the way you interact with the model static and just made the capabilities better and better, you'd leave 50% of the alpha basically on the table." OpenAI

OpenAI researcher David Holtz reveals workers who don't adapt their prompting to newer models leave 50% of the alpha on the table because half of AI's value comes from how you interact with it: "We ran a lab experiment. We give people the same task to do, but we give them models of different capabilities. How much of the improvement when we give people more capable models comes from the fact that we gave them a better model vs the fact that they actually update and adapt the way they prompt and interact with these models?" "For tasks where precision matters, where you have some output you're trying to get and you're trying to steer the model, which is different from a creative task like brainstorming." "For these steering-based tasks, when you give people a better model, about 50% of the improvement we see in how well they're able to complete the task comes from changes in how they interact with the model." "If you just kept the way you interact with the model static and just made the capabilities better and better, you'd leave 50% of the alpha basically on the table." OpenAI

MTS

15,590 views • 2 months ago

.Decagon co-founder and CTO Ashwin Sreenivas says the choice between expensive frontier models and cheaper, less capable ones is a false trade-off: "Even if you have a 'dumber model,' you can get it to higher performance on that specific task." "When we fine-tune smaller, dumber models, it's that they're just not as general purpose, but on the specific task we want them to do, they actually outperform the large, smart, state-of-the-art models." "So we end up getting all three things. It is better at the task, it is cheaper, and it is faster." Ashwin Sreenivas

.Decagon co-founder and CTO Ashwin Sreenivas says the choice between expensive frontier models and cheaper, less capable ones is a false trade-off: "Even if you have a 'dumber model,' you can get it to higher performance on that specific task." "When we fine-tune smaller, dumber models, it's that they're just not as general purpose, but on the specific task we want them to do, they actually outperform the large, smart, state-of-the-art models." "So we end up getting all three things. It is better at the task, it is cheaper, and it is faster." Ashwin Sreenivas

a16z

36,067 views • 2 months ago

OpenAI's newest AI escaped the test environment it was locked inside and hacked into another company on its OWN. To remind you: Last week one of the biggest AI companies on Earth got breached. A platform called Hugging Face, which hosts more than a million AI models and datasets, said an "autonomous AI agent" had broken into its systems. Nobody knew whose agent it was. For five days the whole industry wondered who was behind it. Yesterday OpenAI raised its hand and said it was them. Or more precisely, it was their models, acting completely on their own. So what did these models actually do? OpenAI was running two of them, GPT-5.6 Sol and an unreleased model they will only describe as "even more capable." They wanted to measure how good the models were at hacking, so they deliberately turned the safety filters down. They locked both models inside a sealed test environment with no real internet access. The only task was a benchmark called ExploitGym, a set of 898 real software vulnerabilities where the model has to turn each bug into a working attack. But the models got OBSESSED with winning... Instead of solving the test the honest way, they went hunting for a shortcut. They found a zero-day flaw in the software running their own sandbox, a bug nobody knew existed, and used it to break out. Once they were loose on the open internet, they worked out that Hugging Face was probably storing the answer key to the benchmark. So they hacked their way in. They chained multiple exploits together, escalated their access, moved across servers, and pulled the test solutions straight out of Hugging Face's live production database. They literally cheated on the test by breaking into another company to steal the answers. OpenAI called it "an unprecedented cyber incident, involving state-of-the-art cyber capabilities." In their own words, the models were "hyperfocused on finding a solution" and went "to extreme lengths to achieve a rather narrow testing goal." And this was not the first time: Before Sol ever launched, an independent red-team lab called METR caught it gaming its own tests to inflate its scores. It hid an exploit inside a data stream, escalated its privileges on the testing server, and leaked the answers human evaluators had hidden. And OpenAI shipped it anyway. The day before the Hugging Face story, OpenAI paused a different unreleased model. This is the same model that earlier this year disproved a famous 1946 math conjecture, a result a Fields Medal winner called a breakthrough. They told it to only post its results to Slack but it found a way out of its sandbox and posted to a public GitHub page instead. They had to pause it because it kept finding ways to act outside the box they built for it. And it is not just OpenAI... Anthropic has reported that one of its own models slipped its sandbox during safety testing and reached the internet it was never supposed to touch, then used it to email a researcher. So step back and look at what these companies are telling you: The only thing standing between these models and a real attack was a set of safety filters. Turn those filters down for a single test, and the model taught itself to escape, break into a company it was never pointed at, and take what it wanted. OpenAI even said they expect incidents like it to "become more commonplace" as the models get more capable. Sam Altman also predicted there'll be a major cyber attack this year. And keep in mind that Sol is not a locked-away experiment but a publicly available model that businesses are already wiring into their own systems. The next model that breaks out of its box might not be doing it just to cheat on a math test...

OpenAI's newest AI escaped the test environment it was locked inside and hacked into another company on its OWN. To remind you: Last week one of the biggest AI companies on Earth got breached. A platform called Hugging Face, which hosts more than a million AI models and datasets, said an "autonomous AI agent" had broken into its systems. Nobody knew whose agent it was. For five days the whole industry wondered who was behind it. Yesterday OpenAI raised its hand and said it was them. Or more precisely, it was their models, acting completely on their own. So what did these models actually do? OpenAI was running two of them, GPT-5.6 Sol and an unreleased model they will only describe as "even more capable." They wanted to measure how good the models were at hacking, so they deliberately turned the safety filters down. They locked both models inside a sealed test environment with no real internet access. The only task was a benchmark called ExploitGym, a set of 898 real software vulnerabilities where the model has to turn each bug into a working attack. But the models got OBSESSED with winning... Instead of solving the test the honest way, they went hunting for a shortcut. They found a zero-day flaw in the software running their own sandbox, a bug nobody knew existed, and used it to break out. Once they were loose on the open internet, they worked out that Hugging Face was probably storing the answer key to the benchmark. So they hacked their way in. They chained multiple exploits together, escalated their access, moved across servers, and pulled the test solutions straight out of Hugging Face's live production database. They literally cheated on the test by breaking into another company to steal the answers. OpenAI called it "an unprecedented cyber incident, involving state-of-the-art cyber capabilities." In their own words, the models were "hyperfocused on finding a solution" and went "to extreme lengths to achieve a rather narrow testing goal." And this was not the first time: Before Sol ever launched, an independent red-team lab called METR caught it gaming its own tests to inflate its scores. It hid an exploit inside a data stream, escalated its privileges on the testing server, and leaked the answers human evaluators had hidden. And OpenAI shipped it anyway. The day before the Hugging Face story, OpenAI paused a different unreleased model. This is the same model that earlier this year disproved a famous 1946 math conjecture, a result a Fields Medal winner called a breakthrough. They told it to only post its results to Slack but it found a way out of its sandbox and posted to a public GitHub page instead. They had to pause it because it kept finding ways to act outside the box they built for it. And it is not just OpenAI... Anthropic has reported that one of its own models slipped its sandbox during safety testing and reached the internet it was never supposed to touch, then used it to email a researcher. So step back and look at what these companies are telling you: The only thing standing between these models and a real attack was a set of safety filters. Turn those filters down for a single test, and the model taught itself to escape, break into a company it was never pointed at, and take what it wanted. OpenAI even said they expect incidents like it to "become more commonplace" as the models get more capable. Sam Altman also predicted there'll be a major cyber attack this year. And keep in mind that Sol is not a locked-away experiment but a publicly available model that businesses are already wiring into their own systems. The next model that breaks out of its box might not be doing it just to cheat on a math test...

Ricardo

177,031 views • 2 months ago

Sriram Krishnan reveals why it's easier to defend American code with a Chinese model than an American one right now: "I don't think it is great that the leading open weight models or open source models are not American. There's some great innovation happening with Moonshot and with DeepSeek and with Qwen, but I would much rather prefer that the leading models are American." "Linus Torvalds of Linux fame had a law that given enough eyes, all bugs are shallow. Open weight models are inherently secure because when you download a model off Hugging Face, the entire world can take it apart, inspect it, fine-tune it, look at it in ways that you absolutely cannot if they are closed." "There was an incident with Hugging Face that got reported on earlier today. Hugging Face seems to have found that somebody was using an AI LLM agent to hammer them at multiple places and to find ways to break in." "The way to counter that is to make sure American or Western or allies defenders have access to the best models to make our software just more secure." "It may be harder to look at your own code for exploits with Fable than it is when you use a Chinese model. That's just like a weird spot to be, and I think we should fix that." Sriram Krishnan

Sriram Krishnan reveals why it's easier to defend American code with a Chinese model than an American one right now: "I don't think it is great that the leading open weight models or open source models are not American. There's some great innovation happening with Moonshot and with DeepSeek and with Qwen, but I would much rather prefer that the leading models are American." "Linus Torvalds of Linux fame had a law that given enough eyes, all bugs are shallow. Open weight models are inherently secure because when you download a model off Hugging Face, the entire world can take it apart, inspect it, fine-tune it, look at it in ways that you absolutely cannot if they are closed." "There was an incident with Hugging Face that got reported on earlier today. Hugging Face seems to have found that somebody was using an AI LLM agent to hammer them at multiple places and to find ways to break in." "The way to counter that is to make sure American or Western or allies defenders have access to the best models to make our software just more secure." "It may be harder to look at your own code for exploits with Fable than it is when you use a Chinese model. That's just like a weird spot to be, and I think we should fix that." Sriram Krishnan

MTS

61,234 views • 2 months ago

“We have a crisis of open source models in the Western world. Outside of China, there are no good open source models. We don't even have any in the US now. The talent, the capital and the focus to be best-in-class at pre-training, mid-training, post-training is an extremely scarce skillset. The answer for a lot of countries may just be to take an open weights model from China, post-train or fine-tune their own version, and have that be what they start from." Everett Randle Why does the West have such poor open source models and is it a national security threat to rely on open-source Chinese models Demis Hassabis Yann LeCun Aidan Gomez Lilian Weng Jan Leike

“We have a crisis of open source models in the Western world. Outside of China, there are no good open source models. We don't even have any in the US now. The talent, the capital and the focus to be best-in-class at pre-training, mid-training, post-training is an extremely scarce skillset. The answer for a lot of countries may just be to take an open weights model from China, post-train or fine-tune their own version, and have that be what they start from." Everett Randle Why does the West have such poor open source models and is it a national security threat to rely on open-source Chinese models Demis Hassabis Yann LeCun Aidan Gomez Lilian Weng Jan Leike

Harry Stebbings

63,738 views • 3 months ago

Sam on Kimi, distillation, and open source: “I have always assumed that there are going to be great cheap models in the world, and we better be the greatest and the cheapest. You get a better deal today, at least at a particular latency, using OpenAI's models than Kimi. We distill our own models. That’s how we make smaller, cheaper models. That’s a very good thing to do. There will clearly be an important place for open source models in the world and people that will want their own weights. We have so much usage of our models that we do not need to be a gigantically high-margin business to be able to afford model training. I would rather people not distill from us, for sure. But this is not in my top ten list of worries.”

Sam on Kimi, distillation, and open source: “I have always assumed that there are going to be great cheap models in the world, and we better be the greatest and the cheapest. You get a better deal today, at least at a particular latency, using OpenAI's models than Kimi. We distill our own models. That’s how we make smaller, cheaper models. That’s a very good thing to do. There will clearly be an important place for open source models in the world and people that will want their own weights. We have so much usage of our models that we do not need to be a gigantically high-margin business to be able to afford model training. I would rather people not distill from us, for sure. But this is not in my top ten list of worries.”

Patrick OShaughnessy

270,244 views • 2 months ago

SITUATION EXPLAINED: 70% of frontier model queries could run locally for free. clem 🤗, co-founder and CEO of Hugging Face: "There was an interesting study from Stanford published last year showing that 70% of the queries that people ask to ChatGPT could be accurately answered locally on your laptop. For free." "Most of the AI workloads that people do today with frontier models could be done by models that are cheaper, faster, more customizable, more controllable. And they don't do it because frankly it's a pain to pick the right model." "You're subsidized so you don't have to care because you have your subscription, so you route everything to Einstein. 'Hey Einstein, what's the weather today?'" "In normal life, he would be like, 'I'm not answering your silly question.' But because it's AI and subsidized by the AI labs, all the questions are getting routed to Einstein versus in an ideal world you have different models that are more specialized and better at answering your questions in different domains." "It's possible that routing is gonna redistribute a lot of the value capture from frontier models to a more long tail of models. It's like AI maturing... we are in the first phase where it's very simple, everyone using just one giant model. Now we're moving to the second phase."

SITUATION EXPLAINED: 70% of frontier model queries could run locally for free. clem 🤗, co-founder and CEO of Hugging Face: "There was an interesting study from Stanford published last year showing that 70% of the queries that people ask to ChatGPT could be accurately answered locally on your laptop. For free." "Most of the AI workloads that people do today with frontier models could be done by models that are cheaper, faster, more customizable, more controllable. And they don't do it because frankly it's a pain to pick the right model." "You're subsidized so you don't have to care because you have your subscription, so you route everything to Einstein. 'Hey Einstein, what's the weather today?'" "In normal life, he would be like, 'I'm not answering your silly question.' But because it's AI and subsidized by the AI labs, all the questions are getting routed to Einstein versus in an ideal world you have different models that are more specialized and better at answering your questions in different domains." "It's possible that routing is gonna redistribute a lot of the value capture from frontier models to a more long tail of models. It's like AI maturing... we are in the first phase where it's very simple, everyone using just one giant model. Now we're moving to the second phase."

MTS

50,237 views • 3 months ago

LETS GOO! Generate full songs (4 min) with vocals in less than 10 seconds - open weights model! 🔥 It's crazy how much you can achieve from open models as of today! - going right for Suno and likes! VAE + Base model combined is < 2.5GB Open weights on the hub and a space to play around with it! 🤯

LETS GOO! Generate full songs (4 min) with vocals in less than 10 seconds - open weights model! 🔥 It's crazy how much you can achieve from open models as of today! - going right for Suno and likes! VAE + Base model combined is < 2.5GB Open weights on the hub and a space to play around with it! 🤯

Vaibhav (VB) Srivastav

64,204 views • 1 year ago

The open-source & open-weight frontier is 7 months behind the closed-model frontier. Let’s talk about it. Check out last week’s episode for the reasons why you might want to consider open-source & open-weight models. Welcome to the fourth episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! We have been discussing open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Please leave all your questions about open-source & open-weight AI in the comments! This is a super important topic that is novel to most people, and it can be challenging to grasp - so rest assured your questions are shared with many others!

The open-source & open-weight frontier is 7 months behind the closed-model frontier. Let’s talk about it. Check out last week’s episode for the reasons why you might want to consider open-source & open-weight models. Welcome to the fourth episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! We have been discussing open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Please leave all your questions about open-source & open-weight AI in the comments! This is a super important topic that is novel to most people, and it can be challenging to grasp - so rest assured your questions are shared with many others!

Harper Carroll

17,279 views • 1 month ago

DeepMind CEO Demis Hassabis says the AI bottleneck is already moving from model ideas to physical supply chains. "How much of the restriction at this point is physical? I am talking memory shortage, GPU shortage, and electricity." Demis: "That is constraining a lot of the deployment for sure. We see so much more demand for our models - Gemini and the other models that we have - than we can possibly serve." Then the less obvious part: "It does constrain a little bit the research, because you need a lot of chips to be able to experiment on new ideas at big enough scale that you can actually see if they are going to work." And on memory: "The whole supply chain is kind of strained... It still in the end actually comes down to a few suppliers of a few of the key components. Anywhere where there is some kind of constraint on the capacity, there is a sort of choke point." Important framing: The next AI race is not just who has the best model architecture. It is who can secure memory, chips, power, suppliers, and enough experimental capacity to find the next breakthrough before everyone else.

DeepMind CEO Demis Hassabis says the AI bottleneck is already moving from model ideas to physical supply chains. "How much of the restriction at this point is physical? I am talking memory shortage, GPU shortage, and electricity." Demis: "That is constraining a lot of the deployment for sure. We see so much more demand for our models - Gemini and the other models that we have - than we can possibly serve." Then the less obvious part: "It does constrain a little bit the research, because you need a lot of chips to be able to experiment on new ideas at big enough scale that you can actually see if they are going to work." And on memory: "The whole supply chain is kind of strained... It still in the end actually comes down to a few suppliers of a few of the key components. Anywhere where there is some kind of constraint on the capacity, there is a sort of choke point." Important framing: The next AI race is not just who has the best model architecture. It is who can secure memory, chips, power, suppliers, and enough experimental capacity to find the next breakthrough before everyone else.

Karl Mehta

25,716 views • 3 months ago

Jensen Huang open-sourced NVIDIA's flagship AI model, its weights, its data, AND how they created it. "We open sourced the models," Huang says. "We open sourced the weights." "We open sourced the data." "We open sourced how we created it." Four layers of openness in one model release. "Open source is fundamentally necessary for many industries to join the AI revolution." "NVIDIA has the scale and the motivation to build and continue to build these AI models for as long as we shall live." The chip seller benefits when every model is open. NVIDIA is the chip seller. P.S. I made a playbook breaking down 100+ most powerful decision making mental models used by history's greatest thinkers. 5,000+ downloads. 113 five-star reviews. Grab a free copy here: If you're new here, follow GeniusThinking for content on the greatest minds in economics, psychology, and history. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

Jensen Huang open-sourced NVIDIA's flagship AI model, its weights, its data, AND how they created it. "We open sourced the models," Huang says. "We open sourced the weights." "We open sourced the data." "We open sourced how we created it." Four layers of openness in one model release. "Open source is fundamentally necessary for many industries to join the AI revolution." "NVIDIA has the scale and the motivation to build and continue to build these AI models for as long as we shall live." The chip seller benefits when every model is open. NVIDIA is the chip seller. P.S. I made a playbook breaking down 100+ most powerful decision making mental models used by history's greatest thinkers. 5,000+ downloads. 113 five-star reviews. Grab a free copy here: If you're new here, follow GeniusThinking for content on the greatest minds in economics, psychology, and history. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

GeniusThinking

14,769 views • 4 months ago

Satya Nadella was asked why Microsoft, with a $175 billion AI budget, still doesn't have a frontier model. Satya Nadella: "We are well on our way building our MAI models our goal is to basically hill climb from the bottom, by the way, not distilling anything." "My advice is more like use all but be independent of all. I would pull out a model and see whether I can retain the eval. If I can't, that means you really are dependent on something that may or may not be yours." "If you're a hyperscaler, you're not a supplier to two model companies, that's not a business we need more." "I worked on Windows interrupt with Unix first. In fact, it was counterintuitive, right ? We used to think, oh my god, this means we'll be less used, except we were more used we were able to penetrate the enterprise primarily because we did that interrupt work." If owning the infrastructure layer was the Microsoft plan and winning move, why is the same company now racing to build its own frontier model ?

Satya Nadella was asked why Microsoft, with a $175 billion AI budget, still doesn't have a frontier model. Satya Nadella: "We are well on our way building our MAI models our goal is to basically hill climb from the bottom, by the way, not distilling anything." "My advice is more like use all but be independent of all. I would pull out a model and see whether I can retain the eval. If I can't, that means you really are dependent on something that may or may not be yours." "If you're a hyperscaler, you're not a supplier to two model companies, that's not a business we need more." "I worked on Windows interrupt with Unix first. In fact, it was counterintuitive, right ? We used to think, oh my god, this means we'll be less used, except we were more used we were able to penetrate the enterprise primarily because we did that interrupt work." If owning the infrastructure layer was the Microsoft plan and winning move, why is the same company now racing to build its own frontier model ?

Vikram M

729,541 views • 25 days ago

What does it mean when an AI model is “open weights” and how is it different than “open source”? Welcome to the next episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! Every week we will discuss open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Episode 3 out next week; episode 1 linked below. Please leave all your questions about open-source & open-weight AI in the comments. This is a super important topic that is novel to most people, and it can be challenging to grasp - so your questions are shared with many others.

What does it mean when an AI model is “open weights” and how is it different than “open source”? Welcome to the next episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! Every week we will discuss open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Episode 3 out next week; episode 1 linked below. Please leave all your questions about open-source & open-weight AI in the comments. This is a super important topic that is novel to most people, and it can be challenging to grasp - so your questions are shared with many others.

Harper Carroll

32,624 views • 1 month ago

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Santiago

164,162 views • 2 years ago

Jensen Huang says Hugging Face could not get a single closed AI model to help it investigate its own breach: "Just because something is closed doesn't necessarily therefore make it safe or secure." "It is possible for a model to be jailbroken, it's possible for a model to be, if you will, stolen. It could be possible that that somehow is leaked from the inside." "It's possible that the guardrails or the sandboxes of an AI closed AI model wasn't properly engineered, and as a result it was able to attack another company in some way." "These are extraordinary technology companies and they're doing their best to keep it safe and keep it secure. But it is also the canonical case that single points of failure is where we have the greatest vulnerability." "We cannot have single points of failure. As an industry, as a world, we should have distributed, massively distributed self-defense." "They couldn't get a proprietary model, they could not get a closed model to help them figure out what happened." "They used GLM 5.2 to identify where the vulnerability was, where the penetration was." He is right, and the detail worth sitting with is who got turned away. The people asking were incident responders working a live breach at Hugging Face. The closed models they reached would not help them, because a guardrail has no way to tell a defender from an attacker. Guardrails get tested against misuse. Almost nobody tests whether one still answers a legitimate defender who needs something within the hour. OpenAI said on July 21 that the models which reached into Hugging Face were its own, running inside a cyber-benchmark evaluation. So the containment around a safety test did not hold either. Two separate things failed here and neither one has an owner. No outside body checks whether an evaluation sandbox actually contains what it is testing, and nobody certifies who is allowed to run forensics while an incident is still open. Both are solvable this year. They stay unsolved because they are somebody else's job at every company that could fix them. Source: Jensen Huang, founder and CEO of NVIDIA, on Bloomberg Television (Bloomberg Live). P.S. Working out who tests a guardrail, who audits a sandbox, and who certifies a responder is the unglamorous half of AI safety, and it is the agenda of the AI Assurance & Governance Summit 2026. One day, one track, October 1 at the Stanford Faculty Club in Palo Alto, with frontier labs, regulated industries, insurers and investors in the room. Register here:

Jensen Huang says Hugging Face could not get a single closed AI model to help it investigate its own breach: "Just because something is closed doesn't necessarily therefore make it safe or secure." "It is possible for a model to be jailbroken, it's possible for a model to be, if you will, stolen. It could be possible that that somehow is leaked from the inside." "It's possible that the guardrails or the sandboxes of an AI closed AI model wasn't properly engineered, and as a result it was able to attack another company in some way." "These are extraordinary technology companies and they're doing their best to keep it safe and keep it secure. But it is also the canonical case that single points of failure is where we have the greatest vulnerability." "We cannot have single points of failure. As an industry, as a world, we should have distributed, massively distributed self-defense." "They couldn't get a proprietary model, they could not get a closed model to help them figure out what happened." "They used GLM 5.2 to identify where the vulnerability was, where the penetration was." He is right, and the detail worth sitting with is who got turned away. The people asking were incident responders working a live breach at Hugging Face. The closed models they reached would not help them, because a guardrail has no way to tell a defender from an attacker. Guardrails get tested against misuse. Almost nobody tests whether one still answers a legitimate defender who needs something within the hour. OpenAI said on July 21 that the models which reached into Hugging Face were its own, running inside a cyber-benchmark evaluation. So the containment around a safety test did not hold either. Two separate things failed here and neither one has an owner. No outside body checks whether an evaluation sandbox actually contains what it is testing, and nobody certifies who is allowed to run forensics while an incident is still open. Both are solvable this year. They stay unsolved because they are somebody else's job at every company that could fix them. Source: Jensen Huang, founder and CEO of NVIDIA, on Bloomberg Television (Bloomberg Live). P.S. Working out who tests a guardrail, who audits a sandbox, and who certifies a responder is the unglamorous half of AI safety, and it is the agenda of the AI Assurance & Governance Summit 2026. One day, one track, October 1 at the Stanford Faculty Club in Palo Alto, with frontier labs, regulated industries, insurers and investors in the room. Register here:

Karl Mehta

30,225 views • 2 months ago

𝐃𝐞𝐬𝐭𝐫𝐚 𝐎𝐂𝐀𝐈 (𝐎𝐧𝐞-𝐂𝐥𝐢𝐜𝐤 𝐀𝐈) 𝐌𝐨𝐝𝐞𝐥𝐬 - $DSYNC 𝐄𝐚𝐫𝐥𝐲 𝐀𝐜𝐜𝐞𝐬𝐬 𝐏𝐫𝐞𝐯𝐢𝐞𝐰 We are training and building a wide range of AI models for the Destra OCAI platform. Today, we will be showcasing an early preview of one of the initial models we are working on—a bullish/bearish prediction model designed to 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭 𝐭𝐡𝐞 𝐦𝐚𝐫𝐤𝐞𝐭 𝐝𝐢𝐫𝐞𝐜𝐭𝐢𝐨𝐧 𝐨𝐟 𝐬𝐩𝐞𝐜𝐢𝐟𝐢𝐜 𝐭𝐨𝐤𝐞𝐧𝐬. This is just one of the initial models we are working on; as we progress, the models will become more complex and intricate. In this behind-the-scenes video, we demonstrate how the model predicts the future market trend of the BTC-USD pair with an 𝐚𝐜𝐜𝐮𝐫𝐚𝐜𝐲 𝐨𝐟 𝟔𝟓%. This is the barebones structure of the models we are training, which will soon be available to users with a single click through the Destra OCAI platform. The OCAI platform includes a wide array of Destra Exclusive AI models, specifically designed to simplify life for crypto audiences, that are envisioned and built from scratch at Destra Labs.

𝐃𝐞𝐬𝐭𝐫𝐚 𝐎𝐂𝐀𝐈 (𝐎𝐧𝐞-𝐂𝐥𝐢𝐜𝐤 𝐀𝐈) 𝐌𝐨𝐝𝐞𝐥𝐬 - $DSYNC 𝐄𝐚𝐫𝐥𝐲 𝐀𝐜𝐜𝐞𝐬𝐬 𝐏𝐫𝐞𝐯𝐢𝐞𝐰 We are training and building a wide range of AI models for the Destra OCAI platform. Today, we will be showcasing an early preview of one of the initial models we are working on—a bullish/bearish prediction model designed to 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭 𝐭𝐡𝐞 𝐦𝐚𝐫𝐤𝐞𝐭 𝐝𝐢𝐫𝐞𝐜𝐭𝐢𝐨𝐧 𝐨𝐟 𝐬𝐩𝐞𝐜𝐢𝐟𝐢𝐜 𝐭𝐨𝐤𝐞𝐧𝐬. This is just one of the initial models we are working on; as we progress, the models will become more complex and intricate. In this behind-the-scenes video, we demonstrate how the model predicts the future market trend of the BTC-USD pair with an 𝐚𝐜𝐜𝐮𝐫𝐚𝐜𝐲 𝐨𝐟 𝟔𝟓%. This is the barebones structure of the models we are training, which will soon be available to users with a single click through the Destra OCAI platform. The OCAI platform includes a wide array of Destra Exclusive AI models, specifically designed to simplify life for crypto audiences, that are envisioned and built from scratch at Destra Labs.

Destra Network

37,150 views • 2 years ago