The Quantum Threat to Blockchains - 2026 Report Now available ⬇️

The latest from the Project Eleven blog by @furkanakal Almost everything a blockchain does comes down to checking signatures. Balances, transactions, consensus votes... Strip away the machinery and a node spends its day asking one question over and over. Does this signature verify under that public key? There's no encryption protecting your coins. What protects them is that only you can produce a valid signature for your key. Today, that is done by four schemes: ECDSA, Schnorr, EdDSA, and BLS. They look like four independent choices, but as we'll see, they're really four versions of the same bet. This post kicks off our post-quantum signing series, and the goal is to build the foundation: what a digital signature actually is, how each of the four works, what they cost, and why a single quantum algorithm eventually retires all of them at once. Continue reading: projecteleven.com/blog/post-…

Quantum & Privacy Day 🫡🔥🚀 Apply to attend: luma.com/jtof4o9b

👀👀👀

Proud to support Quantum & Privacy Day with @QuantusNetwork and @NEARProtocol 🫡 Apply to attend: luma.com/jtof4o9b

We’re excited to welcome @projecteleven to Quantum & Privacy Day this October. Project Eleven are leaders building post-quantum security and migration infrastructure for digital assets. October 8 | Tower Club, Singapore Apply to attend: luma.com/jtof4o9b

More progress on quantum means we need more progress on making crypto post-quantum.

Excited for this one! Quantum & Privacy Day during Token2049 Singapore with @QuantusNetwork and @NEARProtocol 🤝

One of the coolest open research efforts in crypto right now. ECDSA.fail is showing what happens when you turn quantum cryptanalysis into a public, competitive research problem. Great work by @eigenlabs/@yukonresearch!

2028 timeframe. 26 days for each 256 bit break. Let's get to work.

We reviewed the reference implementation of Threshold ML-DSA against the paper and found that a requirement the security model enforces had no counterpart in the code: the state of a signing attempt must be used at most once. The API allowed the same round-one randomness to produce responses under two different challenges. Two responses from the same randomness subtract to z − z′ = (c − c′) · s an equation in the party's secret share. This is analogous to nonce reuse in Schnorr. We opened a PR, and after discussion with Guilhem Niot at PQShield on the right design, the fix was merged on September 5. Signing state now moves forward one round at a time, is consumed on use, and the randomness is zeroized. The repository is an academic proof of concept, but for a new construction the reference code ends up being the practical spec. This one has already been reimplemented in other languages. We have since spoken to those teams and helped audit their state handling. Full write up by yours truly: projecteleven.com/blog/key-r…

We found a key-recovery flaw in the reference implementation of the first fully ML-DSA-compatible threshold signature scheme. Reusing signing state could expose a signer’s secret share and, in some settings, the aggregate signing secret. We worked with @PqShield to fix it before the pattern propagated further. @conordeegan explains what we found, why it matters, and how it was fixed: projecteleven.com/blog/key-r…

so the Liquid hackers returned 3,400 BTC and kept 598.5 BTC (~$47m). They’ve called themselves white hats throughout this whole saga. Maybe Blockstream agreed to them keeping the 15%, maybe they didn’t. We don’t know yet. What’s arguably more interesting is what “white hat” hacking really means as AI makes serious vulnerability discovery accessible to a much larger group of people. Historically, the people capable of finding a subtle bug in something like Elements were usually security researchers. There are established norms around what you do next: disclose it, don’t take the money, agree a bounty, etc. AI increasingly separates the capability from those norms and I think we’ve probably underestimated how much security has relied on the two coming together. To get good enough to find really hard bugs, you generally spent years inside security, cryptography or open-source communities. Along the way you didn’t just learn how to find them. You also picked up the norms, built a reputation you cared about, and had professional relationships you probably didn’t want to burn. In other words, responsible disclosure hasn’t only worked because we’ve built good incentives for researchers (arguably most of the time the monetary incentives are crap or non-existent). It’s also worked because the population capable of finding these bugs was unusually selected. AI breaks that selection effect. If the capability becomes cheap and widely available, the norms don’t automatically come with it. That’s a pretty fundamental change to a security model which, it turns out, relied on social infrastructure more than we probably realised.

🚨👀 The Liquid Network hackers appear to have sent 3,400 BTC to Blockstream, keeping ~600 BTC for themselves Is 600 BTC ($50 million USD) a fair white hat reward?

Let’s dive in to the Liquid Network hack. Set a reminder to join! nitter.net/i/spaces/1nJOLQkmjpwxR

Crypto. AI. Quantum. The next decade belongs to all three.