Pinned Tweet
This is Gennady. Gennady holds the upgrade key to your smart contract.
An AI agent wrote a patch at 3 a.m. Gennady read the title and said: ship it.
So we replaced Gennady with a window. Proofs only.
Built at NEAR HACKATHON. @NEARProtocol #NEARHackathon
Emet: a Lean proof checker inside a NEAR contract. No proof, no upgrade. On testnet a gate upgraded real NEAR Intents code only after the chain checked a proof.
Made with @mbor1sov.
Try it: emet.formalized.si/
Code: github.com/formalized-si/eme…
Mike Murunov retweeted
I've been filling in a lot of forms with computer use lately and wanted to look into its refusals.
Sonnet 5.5 and GPT-6.1 Sol always tick "I'm not a robot".
"I typed these answers myself": 0 of 30.
The bot check that works is their honesty, until someone tells them to lie.
Mike Murunov retweeted
Code and every check: github.com/formalized-si/exi…
Illia's Milk Road interview, 18:27: youtube.com/watch?v=xpou72TA…
Which contract should we prove next?
@NearLegion @NearBuilders
Mike Murunov retweeted
Formally verified on mainnet code: the core staking pool, wNEAR and the NEAR Intents token ledger, which never credits more than it holds.
Each model is replayed against the deployed wasm: 311 697 Intents receipts, 0 disagreements. Certificates sit in a keyless mainnet registry.
Mike Murunov retweeted
Every NEAR contract gets an exit stage:
0. its code can change at once
1. only after a visible delay
2. a Lean theorem proves you can always withdraw, tied to the deployed code hash
86.2% of staked NEAR is at stage 2. 22 of 28 large contracts are at stage 0.
Try any account: exit-proof.formalized.si
Mike Murunov retweeted
“If I put money into the smart contract, I can withdraw it back, right? … which by the way requires that the smart contract is not upgradable” — @ilblackdragon on Milk Road podcast
Exit Proof proves it in Lean. With @mike_genproof
Built at NEAR HACKATHON @NEARProtocol #NEARHackathon
The same checker on testnet took NEAR Proof Arena's Lean spec of nearcore's transfers, by @ilblackdragon, as it is. It accepted two of the Arena's own theorems in one transaction each (395 TGas). We prove nothing about nearcore.
1
13
The NEAR Intents team found and fixed the refund bug. Emet found nothing new: their code is the test subject.
A question for builders: which sentence in the comments of your contract belongs in front of its upgrade key?
@NearLegion @NearBuilders
15
Mike Murunov retweeted
New paper: we generate Lean-grounded synthetic data to train verifiers that can find mistakes in natural language math proofs.
Broadly, I hope we can use Lean and formal-to-informal generalization as part of the solution to supervising superhuman models in the future.
bro is cooking
@mbor1sov open question is whether that enables targeted, generalizable edits rather than local overfitting to the evaluation signal (reward-hack4all smelling) + global opt problems (◞ ‸ ◟ㆀ)
Replying to @mbor1sov
Would love to see a benchmark built around this — LLMs reading and writing weights directly. Feels like there's a hidden gem here (new attention is all u need 🤔), and a benchmark could bring more attention to it and help find new methods.
rikiyat.github.io/blog/weigh…
1
68
Mike Murunov retweeted
Lean-eval is the main math formalization benchmark, and we're #1.
We’re releasing our winning harness and sharing our learnings:
genproof.com/blog/quickmaths…
Mike Murunov retweeted
Seems simple? Well, this method is exactly how we found multiple bugs and CVEs in the libraries that Anthropic said they “formally verified”.
See the progress of our bug hunt:
bugs.genproof.com/launch
Mike Murunov retweeted
Consider a compression library. To prove it doesn't corrupt data, you might want this property to hold:
unzip(zip(data)) == data
An LLM generates a proof that this property holds for every valid data input.
The proof is machine-checked, so the LLM output is never trusted.
Mike Murunov retweeted
Formal verification is the ultimate defense against cyberattacks and bugs.
However, until now it required very specialized knowledge.
We’re releasing QuickProof. Now, if you know how to write unit tests, you can formally verify your code.
genproof.com/blog/quickproof
Mike Murunov retweeted
Today, we're launching @genproof: a research lab making formal methods accessible.
We found bugs in 100+ OSS projects, including 6 libraries “formally verified” by Anthropic. We’re also top-1 on the lean-eval leaderboard.
With our tools, you too can become a verifier: 🧵
"another day, another break" - new reality 😭😭😭
half a year was meme "another day -> another rewritten tool in Rust"
1
71
Mike Murunov retweeted
We found two bugs in @deepseek_ai’s DeepGEMM-Ascend, @zheanxu.
One can cause out-of-bounds reads when following the layout docs; the other can silently add the result to the wrong matrix.
Found using AI + FV + property-based testing. 🧙♀️







