
Bas Westerbaan retweeted
CLAUDE FOUND THE SEEDS
anthropic.com/research/ecdsa…
Bas Westerbaan retweeted
CLAUDE FOUND THE SEEDS
anthropic.com/research/ecdsa…
Bas Westerbaan retweeted
Can you build blind signatures from ML-DSA? Yes of course you can! But what will the performance be?
In recent work, we looked at the efficiency of Non-Interactive Zero-Knowledge proofs of knowledge of an ML-DSA signature.
Bas Westerbaan retweeted
yikes: via HDNL 1 Shor "only" gets to you to decrypt one secure channel session, whereas 1 Shor'd PKI cert or auth cert/token can get you access to a whole network/org (modulo the detection heuristics and other auth factors etc.)
Replying to @exocognosis
First quantum attack is expensive and will not be HNDL. Instead it will target powerful widely trusted access keys. If you only forgot to roll one access key you’re in trouble.
Indeed. There is a best algorithm to break RSA-2048 that only God knows. It might be classical or quantum. The quantum one is practical. We hope the classical one isn't. AI helps us answer that question faster. AI doesn't break cryptography: it lets us know it was broken faster.
"discrete log is just like a math conjecture, it's a matter of time and tokens"
No. AI may optimize the best known algorithms and/or implementations, but if factoring/DL are hard (although not NP-complete, if P!=NP we know such hard "NP-intermediate" exist), then no efficient solver exists, ergo no AI can find one. Most likely P!=NP.
< insert caveats about worst/average-case hardness>
Bas Westerbaan retweeted
This is what I have been telling everyone who will listen.
The most valuable thing the frontier labs can now do for crypto (and math more broadly) is publish their *negative* results. What ideas they tried, how much compute they used, and what obstacles they found.
Replying to @matthew_d_green
And specifically for cryptanalysts who are working with models and for the labs:
Folks, what we need right now are *negative* results. Silence is the worst thing you could be publishing.
Bas Westerbaan retweeted
new Keccak variant: MiniSHAKE
800-bit permutation (instead of 1600), only 12 rounds (instead of 24), great idea!
eprint.iacr.org/2026/2363
Bas Westerbaan retweeted
Replying to @WatsonLadd
I'm pretty sure that if have AI and even half-competent people using it -- with high likelihood you will find exploitable issues in the "Type A" equivalents of any country. Good news if your intelligence op is a bit short-staffed on the "top-secret cleared math genius" dept..
OpenSSH 10.6 enables support for post-quantum keys (authentication.) openssh.org/releasenotes.htm…
Zeroization, part 1: Wiping can make things worse 00f.net/2026/10/06/zeroizati…
Bas Westerbaan retweeted
arxiv.org/abs/2610.03682 shows it takes very little non-locality to get big improvements in the overhead of quantum fault tolerance.
If a planar machine is split into pieces, simply changing the piece-to-piece connections can give an order of magnitude improvement in density!
Bas Westerbaan retweeted
i guess we trainin models now
introducing 𝚌𝚕𝚎𝚏: our first models trained by @cloudflare's workers ai team.
today, we're releasing two fast and accurate decision models that top the benchmarks for quality and latency. use them hosted on workers ai or grab the weights from @huggingface, because we open-sourced it too.
blog.cloudflare.com/clef-dec…
Bas Westerbaan retweeted
Cloudflare Workers is adding opt-in support for post-quantum-resistant algorithms ML-KEM and ML-DSA. You can try it today. cfl.re/4hpnCfp #BirthdayWeek
Bas Westerbaan retweeted
A couple of my favorite images from the post-quantum blogs today:
We’re building CryptoLabe, an internal AI-powered tool that discovers cryptography across our codebase, surfaces dependencies, and helps us progress toward a full post-quantum migration by 2029. Here’s what we’ve learned so far. cfl.re/4jeczqX #BirthdayWeek





















