@tysn_dev

Tech + Local AI Own your weights ⚖️ 4x GB10 - 4x AMD MI50 - 1x 3090ti

Joined May 2026
  • Tweets1.3K
  • Following186
  • Followers250
121Media
I remember the anti-AI Reddit cope at the start of vibe coding “in 12 months they’ll scramble to hire senior devs to fix the mess!!” Well turns out you’re the one that left a mess that we can now fix
Anthropic
@AnthropicAI22h
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…

LLMs are making formal software verification feasible at a much larger scale

Claude formally proved security properties in the compiled code of 123 open-source libraries. Along the way, it found bugs that had eluded detection for years.

anthropic.com
16
53
5
1,653
30,121
Ty retweeted
0xSero
@0xSero58m
A bit late but good that you realised it.
Linus ✦ Ekenstam
@LinusEkenstam1h
Time to invest in a small GPU cluster and setup my own completely sovereign compute. Completely isolated in a Tailscale environment, with a cloudflare tunnel. This is nuts, Fable 5.1 capable open model, weights will ship on November 4th. Mark your calendars, take a sick day
4
5
48
2,787
My estimate is 99.999999% of time spent by the average developer right now is on prs and meetings that do not need to exist
3
2
61
799
we’re about to discover that 20 years of software engineering was largely an exercise in accidental complexity, architectural overfitting, and institutionalized mediocrity. most developers optimized for shipping code, not understanding computation.
26
35
13
556
16,729
Can we all admit now that Astra was crazy overhyped? We should treat every release with skepticism until there’s been 2+ weeks of using it.
1
4
135
I’ve realized people with malice in their hearts cannot be convinced, converted, or reasoned with. No matter how hard you try. Love from afar but protect your peace.
18
Why is Muse spark open so much worse than the closed version. Unbelievably disappointing.
1
86
Everyone: "Vibe coding is causing all these security bugs" Reality: PHP, Node, Linux kernel, GitLab. All hand-written code by humans, some of it 20 years old. AI isn't creating the slop. It's finding it 🫠
81
144
18
2,227
39,265
The only thing that stops a bad guy with an abliterated LLM is a good guy with an abliterated LLM.
35
106
26
1,032
20,208
Replying to @NVIDIAAI
@NVIDIAAI GB300 🔥
21
3
194
6,989
I’m getting clowned on for buying these AMD MI50’s for $400 when they are now at $750? Is everyone on Instagram and TikTok lobotomized? I’m genuinely curious. I WILL succeed. I WILL run flash models near or above 100tok/s. And I WILL learn and have fun doing it.
24
4
170
10,063
Absolutely fascinated by this computer stuff. I time travel each time.
1
1
43
Inside the Mikrotik CRS812. The heat sink on the 400G QSFPs (top right) is much larger than the 200G QSFPs for obvious reasons. Don’t get triggered at the click, I was filming with one hand it’s totally fine 😅
2
4
26
2,028
😭😭
dan
@irl_danB17h
Ronald Reagan eats the same Big Mac as you Tim Cook uses the same iPhone as you the NSA runs the same abliterated GLM-5.3 as you
1
3
114
14 year old Dells are still being used in prod Hopper GPUs aren’t going anywhere for a while
Cody Blakeney
@code_starOct 22
a year later and I think they may be still running for 20 years the way GPU rental prices are going.
4
2
31
1,143
I told Deepseek V4.1 Flash 1017 to build a exploit that escapes Chrome's V8 Sandbox After 12 hours, it had a multi stage memory corruption exploit that achieves full process read and write.
45
74
11
2,035
114,668
Load more