Freedom maximalist. Good vibes only.

Boulder, Colorado, USA, Earth
Joined March 2008
Announcing hashsma.sh! 𝟘 𝟙 ❒ 🔨 Strengthen cryptography with AI. Cryptography—both current and post-quantum cryptography—depends on hashes. SHA-256? SHA-3? BLAKE3? Nobody knows if they are secure. We're going to find out! Come help accelerate scientific research! HashSmash is an open science-acceleration project in collaboration with @yukonresearch, @eigenlabs, @projecteleven, and @ShieldedLabs.
61
162
72
778
213,585
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @cyberaxe @zooko
Even with access to the Daybreak model, having purchased a YubiKey, subscribed to the $500 plan, and configured everything correctly @OpenAI continues to impose restrictions on me. I am reaching out to Hey @thsottiaux, please grant all researchers at @yukonresearch / hashmash the freedom necessary to continue their work.
2
2
1
8
676
zooko🛡🦓🦓🦓 ⓩ retweeted
Vibe coded my parents' TV this afternoon! It was showing a sponsored ad before you could even get to Netflix. Gave Claude Code the model number, it found the TV on wifi, removed 25 Xiaomi packages and wrote a new launcher. Looks like an Apple TV now lol
588
854
314
23,609
3,591,852
zooko🛡🦓🦓🦓 ⓩ retweeted
You either die a villain or you live long enough to see yourself become the hero.
Replying to @_pgauthier @Ledger
I'm not sure MtGox has anything to do with that. You received multiple reports of implants in early 2026, and by March you published an article that showed both the hand-made style of implants Joe Grand saw (with the improvised LTE antenna) and the high budget mass produced self-aligning implant. As a security company I have no doubt everyone at Ledger understands what that means: someone is serious about attacking the product. This is a five-alarm fire in the making and without an immediate and appropriate response it could turn into something really bad. So there was a response, but I doubt anyone will find it reasonably appropriate. It does one thing however: it documents Ledger knew about this advanced type of implant in March - the picture is on your site. And if you really think I have no standing talking about this because of MtGox let me add this. I was hoping you guys were better than that. I've been holding Ledger at a higher standard than what happened at MtGox, because I trusted you to do better by learning of my failure - to be better than me. And I'm pretty sure everyone in the industry, especially your customers, have been considering Ledger to be at the very top of the industry in terms of security and disclosure policies and trusting you accordingly. Now I am not sure what to think, not only because of the lack of communication at the time but also because of the continuing downplaying of this incident. I did notice some positive, such as better involvement of the Ledger X account, but also I've seen a lot of people complain they are not receiving responses and you keep calling this the CryptoBilis incident despite similar implants surfacing at other locations (there might be a good reason for that, but a end user doesn't know that, and shouldn't have to worry if their device is altered or not).
1
3
422
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @AndrewCurran_
"Specific population" is the opposite of "universal" in UBI. If they are planning a UBI, they would say "all population". Since they said "specific population", they are planning for something more like "Affirmative Action" funds for "specific population".
3
3
1
90
2,809
zooko🛡🦓🦓🦓 ⓩ retweeted
I've mentioned this pledge that the Anthropic founders took to give away 80% of their equity a few times. People are usually skeptical. I know it's difficult to believe, but for the Anth leadership, immense personal wealth is not their primary motivation.
10
6
1
201
8,078
zooko🛡🦓🦓🦓 ⓩ retweeted
A new contender has joined the fray on hashsma.sh .
8
8
1
95
9,128
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @oconnor663
@oconnor663 @veorq @zooko I wanted a faster go (well, y'know, mostly asm) blake3, so I made 0u.io/b3 - afaik, on desktop zen 5 (using avx512) it's the fastest blake3 single-thread score anywhere (on my desktop, 17GB/s vs 12GB/s for c/rust versions). the trick I'm using there is most applicable to zen 5, but in my testing gave ~10% improvements on modern intel as well (granite rapids). so even though my intel numbers are a little behind rust/c, might be worthwhile taking a look at for that too. also I think generally for all platforms the fastest XOF implementation (in terms of throughput; if you're trying to just pull a random 64 bytes out of somewhere, rust/c handily beat me, to be fair. :)
4
1
4
1,262
Haha! I asked my AI what would the greatest statistician, living or dead, say about our statistical analysis of these benchmarks. My AIs answer: Fisher said consulting a statistician after the experiment is a post-mortem. 😂
10
2
1
91
6,061
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @zkDragon
100% - with AI, we were able to squeeze so much out of the proving system that using Blake3 became more efficient than Poseidon2 :)
1
5
25
1,965
zooko🛡🦓🦓🦓 ⓩ retweeted
While you all being lazy and sleeping, I was out researching at 1am we got promoted for 70.7261 in Blake3 Round 2... Leading the boards #HashSmash @yukonresearch @Zooko's Hash Function Research Contest
1
8
27
2,248
13 hours later … 👀👀👀 hashsma.sh
Wake up, rub eyes, check hashsma.sh … WHOA! 😳 That’s very different than yesterday. @cyberaxe is killing it!
3
16
2
118
16,092
A new contender has joined the fray on hashsma.sh .
8
8
1
95
9,128
zooko🛡🦓🦓🦓 ⓩ retweeted
The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.
2,031
1,190
834
16,100
5,158,771
zooko🛡🦓🦓🦓 ⓩ retweeted
We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked The incident is ongoing so we cannot share more details at this point but we are working with Anthropic, OpenAI, and Google to remediate
2
3
7
1,578
zooko🛡🦓🦓🦓 ⓩ retweeted
Reminder: your Telegram DMs are NOT end-to-end encrypted by default. Regular “cloud chats” are stored on Telegram’s servers with encryption keys Telegram controls. That means Telegram can access your contents. Regular group chats have NO end-to-end encryption option, even when the group is set to private. Even Secret Chats use MTProto, Telegram’s custom protocol, without Signal’s per-message key ratcheting or post-quantum key agreement. Telegram’s privacy policy also says it may retain metadata such as your IP address, device information and username history for up to 12 months. For private conversations, use @signalapp.
Here's how exactly Telegram provides user data to U.S. authorities. A letter recently filed in court shows how Telegram transfers that data, such as IP address and phone number. More: 404media.co/heres-what-teleg…
94
59
1
305
33,785
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @sweis @emilianbold
@sweis you may like this result obtained with Claude on sha256 xitter.cf/futuredies/status/2111…
we filed the first complete, machine-verified differential construction for 39 round SHA-256 applying a new technique to ePrint 2026/1120 allowing to take it further than the paper did. just 0.14 bits below current best generic search submission under the same hashsmash cost model. @cyberaxe join forces?
2
10
1,078
zooko🛡🦓🦓🦓 ⓩ retweeted
Also some valid looking candidates were skipped and don’t fail any post-1997 test we tried. Maybe they have a non-public criteria we haven’t discovered yet, like with DES S-Boxes. Or maybe it’s just software idiosyncrasies. I think that’s very likely.
2
2
7
1,176
zooko🛡🦓🦓🦓 ⓩ retweeted
Did you just realize that you left the coin center dinner with the wrong raincoat? Check your closets gals. I have yours and you have mine. Let’s swap back?
4
2
8
3,370
zooko🛡🦓🦓🦓 ⓩ retweeted
Replying to @zooko
🙏I felt that if I post it there's better chance someone optimizes or refutes cause I we've been looking at it alone for too long now
1
2
8
1,030
👀👀👀
we filed the first complete, machine-verified differential construction for 39 round SHA-256 applying a new technique to ePrint 2026/1120 allowing to take it further than the paper did. just 0.14 bits below current best generic search submission under the same hashsmash cost model. @cyberaxe join forces?
5
7
2
102
11,503