ava retweeted
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
636
1,064
9,705
2,026,972
int F[256], k = 0;
for (int i = 0; i < 256; i++) F[i] = -1;
srand48(48571);
while (k < 256) {
int v = 256 * drand48();
if (F[v] < 0) F[v] = k++;
}
for (int i = 0; i < 256; i++)
printf("%02x%c", F[i], i % 16 == 15 ? '\n' : ' ');
for (int i = 0; i < 256; i++) F[i] = -1;
srand48(48571);
while (k < 256) {
int v = 256 * drand48();
if (F[v] < 0) F[v] = k++;
}
for (int i = 0; i < 256; i++)
printf("%02x%c", F[i], i % 16 == 15 ? '\n' : ' ');
Claude and @avhwl also found that Skipjack cipher's s-box is generated by srand48(48571) plus some swaps.
3
3
33
4,215
apparently it is obvious from looking at the bits, if you're a claude.
CLAUDE FOUND THE SEEDS
anthropic.com/research/ecdsa…
3
0
13
525
ava retweeted
multi-asset shielded transactions are live on our testnet, if you want to play with them :)
The Valetudo testnet is the smallest possible testnet, just a single Tendermint instance on a single VPS, but it’s a big step forward, with our MVP1 milestone: multi-asset shielded transactions and a basic, fully-trusting light wallet protocol for client-side chain scanning.
1
6
30
ava retweeted
I don't know any golang devs, I just know a lot of temporarily misguided rustaceans
8
8
110
i'm looking forward to the future of socially acceptable business vtubing
Microsoft Teams enters the metaverse race with 3D avatars and immersive meetings
theverge.com/e/22523015
0
0
1
ava retweeted
wannabe sociologists always pontificate shit like "our innate desire for big numbers drives us toward polarizing content online" and nobody ever considers that maybe some facebook programmer one day was just like "thumbs up = 1; angry face = 5"
‼️#Facebook programmed its algorithm that decides what people see in their news feeds to push content likely to make them ANGRY, treating ANGRY emoji reactions 5x more valuable than “likes.”
Anger kept users more engaged, making #Facebook more money.🤬
washingtonpost.com/technology/202…
2
50
123
ava retweeted
One observation (by @Aelkus) that’s stuck with me is that for all the talk about misinformation, basic information provenance tools like reverse image search have been integrated on 4chan (!) for a decade+ and remain absent from mainstream social platforms.
I’m not ready to concede this point. We have not yet had a platform that helped distinguish untrustworthy sources that did not manipulate its feed for engagement.
2
15
166
ava retweeted
Adding an X under gender on official documents functionally marks people as non-normative and worthy of extra policing. This is an example of reformism making us both visible and more vulnerable—abolition of gender/sex markers on official documents should always be the demand.
13
1,674
7,316
ava retweeted
A common web2 model is where there’s some open-source tool that’s very useful, but can’t internally account for the economics involved in sustaining it, so some company encloses it, and uses the resulting monopoly rents to account for those costs internally.
1
4
28