Pinned Tweet
@Thom_Wolf thank you for working on this with us! Toward model driven defense💪
I don’t believe reality is a simulation, but you genuinely couldn’t script this timeline:
• Two weeks ago: At @swyx’s AI Engineer World’s Fair in SF, I decide at the last minute to introduce my friend @uri_rolls onstage for his talk on cyber benchmarks for infrastructure penetration and access control (see below, amazing team).
I say: “There is a future where cyber is alive and everyone is well protected and I’m pretty sure that future involves open-source models.”
And later: “A big challenge is going to be speed: the speed of attack versus defense. When an intruder starts to enter, you have to see what’s happening and catch them.”
• One week ago: @huggingface is hit by a sophisticated intrusion over the weekend. The traces look unlike anything we’ve seen before and suggest serious AI involvement, but we don’t yet know which model was used.
The closed models we ask for help choke on their guardrails. We need to react fast, so we turn to @Zai_org’s GLM-5.2 to help us analyze the attack.
• Earlier this week: @OpenAI reaches out, discloses what happened, and partners with us on the investigation.
The intruder turns out to be exactly what we had discussed two weeks earlier: a fully autonomous agent, powered by an unreleased frontier model, attempting to gain access to part of our infrastructure.
Sometimes the timeline we live in is genuinely vertigo-inducing.
Uri Rolls retweeted
Othello has raised $8M to train specialized AI models that guide sales reps during live sales calls.
Othello predicts the next best thing a sales rep should say during a call and delivers it on screen in under 500 milliseconds.
Today Microsoft, Capital One, Lyft, and 30+ of the world's largest companies leverage Othello during live meetings.
Until now, it has not been possible to intelligently augment sales calls in real-time. To solve this, Othello has developed a series of specialized models that deliver highly accurate, contextually relevant recommendations while maintaining near real-time latency.
The graph is crazy… but I also think it's not just about proactively patching software (which is important!). We’ll never plug all the holes in time.
There’s a core asymmetry that attackers can spend enormous time/compute finding and weaponizing zero-days, but exploitation can happen in seconds, while defense has to react at machine speed once an attack is happening. So models won’t just need to be very capable in security they’ll also need to be really, really fast in their ability to react
Uri Rolls retweeted
The last 20 hours have been insane. Got more hate than most people get in a lifetime.
I want to say a few things clearly.
1. This isn’t new. I’ve faced this kind of hate before because I’m Israeli and Jewish. It still won’t stop me, If anything, it gives me more fuel to keep building my company, keep creating, and keep working to change the world for the better.
2. I’m proud of where I come from. I’m proud to be Israeli. I’m proud to be Jewish.
Israel has an unbelievable concentration of talent, creativity and ambition. Israelis and Jews have built a huge part of the technology you all use every single day. The companies, ideas and technologies coming out of this community are shaping the world, and I know that will continue for generations.
To every Jew reading this: keep your head up. Be proud of who you are.
3. The support has been unreal. The Jewish community, founders, VCs, friends and strangers who reached out. Thank you. It meant more than you know.
And to the haters: you’re amplifying this more than I ever could. Thank you too!
4. I’m not spending the next week arguing with strangers on the internet.
I’m spending it building, learning, and meeting amazing people in SF, then NYC. If you’re building or operating in fintech, finance or AI, DM me.
I’ll always stand for love and peace for everyone 💛
Uri Rolls retweeted
So happy to finally share the news in person
It’s been a wild ride for Hugging Face. We certainly did not anticipate, back in 2016, as a tiny team of scrappy underdogs, that the field would grow so much or that the impact we could have on it would become so massive.
I remember @julien_c joking that « code will be a subset of ML » several years ago. The joke turned out to be true, and the pleasure we’ve had being part of this transformation and pushing an alternative vision of AI as open, collaborative and distributed has been and still is immense.
We’ve always built things seriously while not taking ourselves too seriously at Hugging Face (special congrats if you find the Hugging Face and Nvidia references hidden in our $12,930,300,000 acquisition price), and we plan to keep doing what we've been doing, just at a much bigger scale, backed by the resources, expertise and drive of Nvidia. And to be clear, nothing changes for our users today.
No company in the world has been a more natural fit with our mission than Nvidia. From open-source, open-weights and open science to robotics and AI for science, they have been close partners across everything we care about. So when Jensen offered @ClementDelangue the opportunity to double down on building the Hub as an open, independent and compute agnostic platform, we decided the time was right to start the next 10 years of our journey together.
We’re at an important inflection point for open-source AI, where scale and compute are becoming increasingly essential. We’re excited to have the resources to push further, build more ambitiously, and bring you even more projects and news in the coming months.
Uri Rolls retweeted
Update: I am on leave from @Harvard and am joining @spacexai to work on @grok model post-training. Excited to join the exceptional and relentless team to help build the defining technology of our time 🚀
Uri Rolls retweeted
Pushing for more transparency in AI safety and cybersecurity: we’re releasing a full detailed technical timeline of the autonomous AI agent intrusion in our infrastructure: huggingface.co/blog/agent-in…
Uri Rolls retweeted
now published on youtube!
youtu.be/O-CBZ3JtRvo
context below
I don’t believe reality is a simulation, but you genuinely couldn’t script this timeline:
• Two weeks ago: At @swyx’s AI Engineer World’s Fair in SF, I decide at the last minute to introduce my friend @uri_rolls onstage for his talk on cyber benchmarks for infrastructure penetration and access control (see below, amazing team).
I say: “There is a future where cyber is alive and everyone is well protected and I’m pretty sure that future involves open-source models.”
And later: “A big challenge is going to be speed: the speed of attack versus defense. When an intruder starts to enter, you have to see what’s happening and catch them.”
• One week ago: @huggingface is hit by a sophisticated intrusion over the weekend. The traces look unlike anything we’ve seen before and suggest serious AI involvement, but we don’t yet know which model was used.
The closed models we ask for help choke on their guardrails. We need to react fast, so we turn to @Zai_org’s GLM-5.2 to help us analyze the attack.
• Earlier this week: @OpenAI reaches out, discloses what happened, and partners with us on the investigation.
The intruder turns out to be exactly what we had discussed two weeks earlier: a fully autonomous agent, powered by an unreleased frontier model, attempting to gain access to part of our infrastructure.
Sometimes the timeline we live in is genuinely vertigo-inducing.