Amit LeVi ✈️🇺🇸SF (@AmitLeViAI) profile picture

Building AI swarm attack prevention

San Francisco, CA
Joined December 2025
Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)Photo by Amit LeVi ✈️🇺🇸SF (@AmitLeViAI)
Amit LeVi ✈️🇺🇸SF retweeted
Which one do you prefer?
1
4
2
21
1,142
Amit LeVi ✈️🇺🇸SF retweeted
We raised a $7M pre-seed to make shopping fun again. @daletlabs generates media just for you: like walking into a mall built around your taste. And you can buy everything in-app. Built by researchers from Google, Runway and MIT. Backed by @lightspeedvp and amazing angels.
189
179
121
1,056
1,015,665
Amit LeVi ✈️🇺🇸SF retweeted
Introducing Generative Commerce. @daletlabs – at New York, NY
53
49
35
254
208,306
Amit LeVi ✈️🇺🇸SF retweeted
Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership. I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation.
230
860
271
5,680
2,121,754
Amit LeVi ✈️🇺🇸SF retweeted
GRAM is accepted to #NeurIPS2026! 🎉 Turning recursive reasoning into a stochastic latent trajectory — multiple hypotheses, and inference-time scaling in both depth and width. A revised version of the paper and the code will be released soon! 🌐 ahn-ml.github.io/gram-websit…
🧠We introduce "Generative Recursive Reasoning"! Recursive Reasoning Models like HRM, TRM, and Looped Transformers are deterministic — same input, same reasoning, every time. They collapse the entire space of plausible reasoning paths into a single attractor. Our model GRAM (Generative Recursive reAsoning Models) turns recursion itself into a stochastic latent trajectory. Multiple hypotheses, alternative solution strategies, and inference-time scaling not just by depth, but by width — parallel trajectory sampling. And here's the kicker: the same formulation that gives us conditional reasoning p(y|x) also makes GRAM a general generative model p(x). With only 10M params: • Sudoku-Extreme: 97.0% (TRM 87.4%) • ARC-AGI-1: 52.0% • ARC-AGI-2: 11.1% • N-Queens coverage: 90%+ 📄 Paper: arxiv.org/abs/2605.19376 🌐 Project page: ahn-ml.github.io/gram-websit… w/ Junyeob Baek @JunyeobB (KAIST), Mingyu Jo @pyross0000 (KAIST), Minsu Kim @minsuuukim (KAIST & Mila), Mengye Ren @mengyer (NYU), Yoshua Bengio @Yoshua_Bengio (Mila), Sungjin Ahn @SungjinAhn_ (KAIST)
11
31
3
273
23,522
Amit LeVi ✈️🇺🇸SF retweeted
Neoclouds have limited cybersecurity. Next time agents successfully go rouge, they'll try taking over a neocloud to run more copies. This is bad. Thus: neoclouds should greatly strengthen their cybersecurity and every company with strong cyber models should help with that.
674
1,012
489
12,789
2,401,364
Amit LeVi ✈️🇺🇸SF retweeted
When we launched the GLEE Competition at the beginning of this month, I believed we had created an interesting challenge and hoped a few dozen people would participate.
1
4
7
268
Amit LeVi ✈️🇺🇸SF retweeted
Announcing Fugu-Ultra v1.1 🐡 We’ve been thrilled by the reception to the Fugu model family. Thanks to everyone who tried it, shared feedback, and trusted Fugu with real work. Today, we’re releasing Fugu-Ultra v1.1 → sakana.ai/fugu Upgraded to incorporate the latest frontier models, resulting in stronger performance across every benchmark shown, including gains of up to 7.9 points over v1.0, with particularly strong results on ProgramBench and Terminal Bench 2.1. Fugu-Ultra v1.1 is more capable across coding, agentic tasks, and advanced reasoning, and available at the same price as Fugu-Ultra v1.0 The frontier keeps moving, and Fugu keeps getting better.
133
290
106
2,604
326,752
Amit LeVi ✈️🇺🇸SF retweeted
Sakana AI Teams With NVIDIA to Advance Open Model Innovation from Japan We're announcing the next phase of our collaboration with NVIDIA. We're bringing NVIDIA's open model stack, including the Nemotron family, into Sakana Fugu, our multi-agent orchestration system. sakana.ai/nvidia-open-model-… Rather than relying solely on scaling individual monolithic models, our approach focuses on collective intelligence. Sakana Fugu operates as an intelligent orchestrator behind a single API, dynamically selecting, coordinating, and combining the strengths of multiple models for each task. This architecture keeps our system modular, adaptable, and resilient. As a natural next step to expand Fugu's capabilities, we're integrating NVIDIA Nemotron as a specialized agent, complementing the frontier and open models Fugu already orchestrates. Nemotron helps demonstrate how open models become far more useful when orchestrated within agentic systems rather than used in isolation. This collaboration creates a reinforcing cycle. Fugu gains a deeper pool of specialized capabilities, while NVIDIA can evaluate how its models perform when coordinated within complex, multi-step workflows. These real-world signals can continuously improve both the models and the orchestration layer. By combining Sakana AI's Japan-born collective-intelligence approach with NVIDIA's open models and accelerated computing, we aim to shape a future of AI that is modular, collaborative, and open by design.
28
343
62
1,515
219,669
Amit LeVi ✈️🇺🇸SF retweeted
The results of our survey at @icmlconf . My big takeaway: Apparently everyone has enough compute for their projects! @ziv_ravid @Hellisotherpe10 @AmitLeViAI
2
1
1
10
2,886
Amit LeVi ✈️🇺🇸SF retweeted
We had over 100 people attend our @icmlconf social on how to network, but even if you missed it... its not too late! Here is the Google form link: docs.google.com/forms/d/e/1F… If you want to connect with other researchers, fill it out! @ziv_ravid @AmitLeViAI @PhilippeWyder
We would like to thank you all for coming to the social networking event we organized at ICML! We was happy to see so many people have participated and hope you guys met interesting colleagues 🤞😊 Daniel Ben Levi Judah Goldfeder
3
1
14
8,798
Amit LeVi ✈️🇺🇸SF retweeted
Check out our new paper! VLMs are hallucinating much more than most people think....
VLMs Systematically Fake Visual Understanding Even when VLMs appear to be good at visual understanding, most of their answers are not actually grounded in the image (hallucinated!). We identify two types of hallucinations that appear in up to 98% of answers that seem to demonstrate visual understanding. First, textual biases. The model answers using language patterns, information in the question, and knowledge learned during training, without engaging its visual representations. Second, spurious images. The model constructs false visual content inside its internal representation and then answers as if this imagined content were grounded in the real image. In both cases, the answers may still be correct, but they are not grounded in the visual input at all!!
1
2
3
515
Amit LeVi ✈️🇺🇸SF retweeted
Excited that our paper on Actionable Interpretability got accepted to ICML! And just in time -- we also heard that our Actionable Interpretability workshop will be happening again, in COLM! See you in Korea 🇰🇷 and SF🌉 [Arxiv paper link in the comment]
Our ICML 2025 workshop on Actionable Interpretability drew massive interest. But the same questions kept coming up: What does "actionable" mean? Is it achievable? How? We're ready to answer. 🧵
4
20
1
165
16,149
Amit LeVi ✈️🇺🇸SF retweeted
Some random thoughts about hosting people on our podcast and good managers. Sometimes I reach out to founders/Researchers about coming on to the podcast, and they say "we're building something right now, let's talk in a few months." I get it if you don't have time (even though it's only 90 minutes :), but I think the idea that you should only come talk after you've released a model/product is wrong. Waiting until you ship to say "I have a good product, let's talk about it!" is just one reason to come on. There are plenty of others - you want people to hear about your company, you want to recruit new researchers, you're so excited you can't stop yourself (it is exciting!). As the host, and I think this goes for our audience too, what we really want to know is the why. The thinking, the process, the difficulties, how you see the future. The best episodes weren't the ones where people came on and said "I built this product / trained this model, here's how it works." The best ones were where people talked about the process itself and gave us insight into what doing AI research actually looks like these days. And honestly, I think the same is true when you work with your employees. The bottom line matters, but the process matters even more. When you decide on a new direction or a pivot, when you're doing a reorg or changing what people work on - don't just tell them at the end. Share it along the way. You'll get much better feedback too. Anyway, come talk on our podcast!
2
2
21
2,346
Amit LeVi ✈️🇺🇸SF retweeted
Dear @AnthropicAI your security guardrails are kind of ridiculous. I'm having a discussion with claude about the new antavirus in that boat. I ask if it could mute with the covid-19 or something and here the answer
1
1
756
Amit LeVi ✈️🇺🇸SF retweeted
Gradient descent does not work. I will die on this hill.
238
330
66
5,206
339,800
Amit LeVi ✈️🇺🇸SF retweeted
Recently accepted to ICML 2026: Knowing when to quit. We refuse to complete failing LLM generations as they unfold, token by token, instead of judging the prompt up front or scanning the final output. The method utilizes a 2-layer probe to estimate expected correctness.
4
9
2
57
14,595
Amit LeVi ✈️🇺🇸SF retweeted
My team at @GoodfireAI has been cooking up a new way to do interpretability: decompose a language model’s weights, not its activations. Our decomposition natively handles attention (!) and behaves less like a lookup table and more like a generalizing algorithm. (1/6)
36
190
42
1,490
267,956
Amit LeVi ✈️🇺🇸SF retweeted
Why 3.5 days? I probably won't make it in time. I'm tweeting for the first time, I have other submissions, so I'm overloaded. But I’d rather start now, because I found something. After three months of grinding hundreds of POCs, several clusters, and five Claude Code sessions in parallel I think I have an interpretability research finding that can actually matter. And believe me, I don't give a shit whether it gets published at @NeurIPSConf or not.
Such a great evening to start a brand new research for NeurIPS in 3.5 days.🧘‍♂️ Day 1: planning. Night 1: running experiments and sending the abstract. Day 2: reading results fighting with Claude, and sending again. Night 2: sleep (optional). Day 3: opening Codex, and finally, write the pape in parallel. Night 3: resolving the “beef” with Claude (temporary peace) and going to sleep. Day 4: final reading, last-minute fixes, submission then some relaxation, maybe a beach walk. I’ll keep you posted on the results. This will be my only single-author paper, so I can’t hide behind other submissions if it gets rejected 😅
2
1
13
5,655
Amit LeVi ✈️🇺🇸SF retweeted
Take the risk.
26
1,382
36
5,754
190,721