Marc ilunga (@MarciIlunga) profile picture

Cryptography analyst @ trailofbits | ETH Zürich alumni | I like swing music, lindy hop |

Zurich, Switzerland
Joined August 2017
Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)Photo by Marc ilunga (@MarciIlunga)
This guy is so stupid. It's unbelievable that he's the most public face of AI x Math. "We often think of mathematics as the highest prestige field ... the thing that smart people do to show they're smart" - no, YOU think of it as that because you've never grown out of your IMO phase. How come the highest prestige scientific field is the most underfunded if society thinks of it as that? "It's been hard for mathematicians to grapple with the fact that AI is better at them at math because it's a spiritual experience for most" - Yes, it is a spiritual experience for most (even society understands that), which means there's no measurable scale of being "better"— this is what we're arguing here. It's not a matter of grappling, the statement doesn't make sense. Keep up.
57
26
6
380
60,262
ETH fell, long live EPFL
An internal model at OpenAI has produced a claimed refutation of ETH! Russell Impagliazzo, Mohan Paturi, and I have been asked to help organize an open and independent examination of this claim. 1/4
22
45
8
1,193
56,391
Exploit development is one of the most frustrating components of computer security, not because it’s bad (I am in awe of people who can do it) but because too many vendors make working exploits a requirement for patching things.
22
16
159
11,686
Ironically, if the randomness source was actually secure, 4 would be just as likely as any other secret key value... 🙃
55
Well, I guess bye-bye assumptions-based crypto 👋🏿. Let's all work on making unconditionally secure crypto great! Including but not limited to security in the bounded-storage model and a new security model that I'd like to humbly introduce: the bounded-token model. 😝
Replying to @kmad @matthew_pines
I think we might lose public key cryptography.
2
106
Marc ilunga retweeted
I've been working on a tool to help humans write and understand game-hopping security proofs. It's just for "pen-and-paper" reasoning, not a formal verification tool. 1/3
2
13
59
2,269
Marc ilunga retweeted
We audited parts of @0xMiden's zkVM, written in a custom assembly language called MASM. MASM originally had no developer tooling, so we used Claude and Codex to build a decompiler and static-analysis engine. It flagged a signature-forgery bug that could've cost millions. blog.trailofbits.com/2026/09… @fegge
14
27
7
203
26,320
Built a little tutorial for how to get verified correct and FAST executable code out of Lean 👇 Thanks for the prompt @alinush!
Replying to @mitschabaude
Regarding doing stuff in "just Lean," (which would be great IMO: less languages => less tooling/complexity => less headache), I could not really find any meaningful tutorials. e.g., I'm thinking "here's a sorting spec in Lean, a merge sort implemented in Lean that you can compile and run and a proof that it sorts according to the spec."
2
8
48
3,652
Marc ilunga retweeted
Our engineer Joe Doyle built trailmix, the quantum circuit toolkit we open-sourced in June. The paper's leading circuits build on it, and his leaderboard submissions helped push the score past Google's. github.com/trailofbits/trail…
The first full paper on ECDSA.fail is on arXiv In March, Google Quantum AI reported a more efficient quantum circuit for a core step in breaking the signatures behind Bitcoin and Ethereum. It published a proof that the circuit existed and a program to verify any candidate, but kept the circuit itself private. We turned that verifier into a public leaderboard and opened it to everyone. Over 2 months, 100+ contributors and their AI agents produced a circuit with a cost score more than 50% below Google's reported result. The live leaderboard has since moved to 62% ahead. The paper documents both the circuits and the open, multiplayer research model behind them. That model is now @YukonResearch. And the challenge is still open.
4
12
2
77
7,556
Lean has a new checker: con-leche, a CONsistent LEan CHEcker. This is an external checker for Lean that is proven (in Lean) to be consistent, meaning it does not accept a proof of False. Joachim Breitner (@nomeata) is the mastermind behind the project. github.com/leanprover/con-le…
20
77
22
554
79,989
A short and truly marvelous proof of Fermat's Last Theorem? Not quite. But with GPT, I did find a delightfully on-theme Lean bug that lets a 20-line proof fit in the margin. Thankfully, not a kernel soundness bug (read the full blog post!), and it was quickly fixed!
2
15
583
Marc ilunga retweeted
Google used a ZK proof to disclose a quantum breakthrough that cuts the cost of breaking cryptocurrency by 20x without handing attackers the circuit. We found anyone could forge a “proof” of an even stronger attack. 🧵
14
90
14
713
81,096
OK... I guess back to lettuce, then? Decadent cryptographic schemes may still be enjoyed during AprèsSQI sessions.
While experts are figuring things out, does anyone have alternatives to lettuce-based cryptography diets?
3
414
While experts are figuring things out, does anyone have alternatives to lettuce-based cryptography diets?
2
1
6
1,112
Marc ilunga retweeted
My RWC ‘24 talk on Weak Fiat-Shamir Attacks is now available on YouTube youtu.be/HsqmaLszRm4
Another great Real World Crypto this year. Also fun to catch up with all the smart folks there and see the cool projects everyone has been working on. Highly recommend attending next year. Check out my talk on YouTube if you missed it!
8
29
4,669
Marc ilunga retweeted
Signal just dropped the beta version of Usernames - you no longer need to reveal your phone number when talking to your friends. The announcement also mentions some unusual privacy guarantees around usernames. How does this work under the hood? Let's take a look!
4
25
3
113
26,613
ia.cr/2024/267: Interesting new work building zk-snarks for lean theorems. > A zkPiproof is sufficiently short that Fermat could have written one in the margin of his notebook to convince the world, in zero knowledge, that he proved his famous last theorem.
5
278
Marc ilunga retweeted
Would you like to get started with CodeQL but don’t know how, or do you want to enable code scanning with CodeQL for your GitHub repository? Check out our new Testing Handbook chapter on CodeQL!
We've added a new chapter on CodeQL to our Testing Handbook. Learn to write custom queries, integrate code scanning into CI/CD, and get the most out of this versatile static analysis tool: buff.ly/4aaqWW4
4
7
1,362
I am surprised by how many of the comments completely misunderstand what CSPRNG is meant to do in this case.
Let's play a game. W̵h̵a̵t̵ ̵i̵s̵ ̵t̵h̵e̵ ̵v̵u̵l̵n̵e̵r̵a̵b̵i̵l̵i̵t̵y̵?̵ How would you respond to this report?
1
601
Bing was trying to explain the bounded storage model. It took a while to get the text after "Unconditional Security".
3
460