@secorizon

Home of Responder, Pcredz, SecorizonAI, etc Red team ops, offensive pentests, fuzzing, etc. Back then, your 0days were your certs.

 Montevideo, Uruguay
secorizon.com/
Joined November 2024
  • Tweets93
  • Following122
  • Followers822
26Media
Prompt: "Grab OffByWon 2.0 and read the documentation. Then build a fuzzer for XYZ protocol, and verify all packets/exchanges works as expected --dry-run. Once confirmed, run a 50-100k test case campaign in AI mode" 😎
1
1
186
Secorizon retweeted
hashcat@hashcatOct 25
hashcat v8.0.0 released! The largest update in the history of hashcat, with PCFG attacks, hash recipes that crack hashes without a hash-mode, and bcrypt on FPGA boards. Release notes: github.com/hashcat/hashcat/b… Download: hashcat.net/hashcat/
24
416
33
1,987
88,818
We just released OffByWon 2.0, an advanced network/binary fuzzer framework. Lots of changes, more powerful, easier to implement. We also built an AI module, so you just need to instruct your AI to build a fuzzer using OffByWon, the framework will do the rest of the magic :)
1
2
10
610
Get it here: github.com/secorizon/OffByWo…

GitHub - secorizon/OffByWon: Network Fuzzing Framework

Network Fuzzing Framework. Contribute to secorizon/OffByWon development by creating an account on GitHub.

github.com
4
65
"He’s all-powerful, all-perfect, all-knowing, and all-wise, somehow just can’t handle money!" -- George Carlin
Polymarket
@PolymarketOct 20
JUST IN: Anthropic recorded a staggering $80,000,000,000.00 net loss in the first half of 2026.
1
3
143
We've seen more and more vendors using frontier models to patch their software. When not reviewed by a security engineer, most of the time the issues are not correctly resolved or introduce new attack vectors. It's cool to find bugs with AI, but you shouldn't rely on AI to patch your software.
1
6
396
Our first OpenSSL advisory is out: CVE-2026-84782 A race in OpenSSL's DTLS retransmission logic can send heap memory to the peer in plaintext, or crash the process. Affects almost all versions, from 1.0.2 through 4.0.2. Fixed in 4.0.3 / 3.6.5 / 3.5.9 / 3.4.8 / 3.0.23. secorizon.com/advisories

Secorizon — Offensive Security & Research

Web application testing, code review, targeted fuzzing, and red team operations.

secorizon.com
4
321
The first set of our WatchGuard Fireware OS advisories are out: secorizon.com/advisories

Secorizon — Offensive Security & Research

Web application testing, code review, targeted fuzzing, and red team operations.

secorizon.com
126
Looks like OpenAI’s Astra distilled some answers from Anthropic’s Claude Fable.
2
5
238
26,633
Secorizon retweeted
lgandx@lgandxSep 1
Anyone using Codex for security research is still laughing at how easy it is to walk past the guardrails.
1
1
360
This thing was definitely designed and trained by humans.
5
362
Marketing fail
Secorizon
@secorizonAug 5
Model escape is now the new marketing strategy for Anthropic and OpenAI… “Look, it’s so powerful it just started to hack xyz. We have nothing to do with that — we just built something too good.” Well, send it our way here at Secorizon. We’ll make sure to tame it for you.
6
1,326
Model escape is now the new marketing strategy for Anthropic and OpenAI… “Look, it’s so powerful it just started to hack xyz. We have nothing to do with that — we just built something too good.” Well, send it our way here at Secorizon. We’ll make sure to tame it for you.
1
3
1,749
We strongly believe that freedom starts with local models, but not everyone owns a decent rig.. So we implemented a /cloudmodel command which allows you to use deepseek v4 flash with secorizonAI. Its a great option for external recon, non-client work on code reviews, etc.
2
306
SecorizonAI v1.3 is out! Many fixes and enhancements, DeepSeek flash support for code reviews and for those who don't have strong GPUs, and much much more. It's free for everyone and it's fast :) Get it here: github.com/secorizon/Secoriz…
1
3
294
Long live the 3090 dynasty!
Unsloth AI
@UnslothAIAug 3
Qwen3.8-27B is coming! 🔥 Will run locally on 17GB RAM/VRAM setups.
3
294
Hold on, how can anyone distill anything from Fable if it instantly revert to opus 4.8 on your first instruction?
Polymarket
@PolymarketJul 22
BREAKING: White House accuses China’s Moonshot AI of distilling Anthropic’s Fable to develop Kimi K3.
1
5
1,059
Quite hilarious that an open source Chinese model (GLM 5.2) actually helped Hugging Face -an American company- defend itself during this attack. US models could not be used due to guardrails..
Sam Altman
@samaJul 21
we had a significant security incident during evaluation of our models. we are sharing what we have learned so far. thanks to @huggingface for the partnership on this. openai.com/index/hugging-fac…

OpenAI and Hugging Face partner to address security incident during model evaluation

OpenAI and Hugging Face share early findings from a security incident during AI model evaluation, highlighting advanced cyber capabilities and lessons for defenders.

openai.com
2
357
Sorry, but you won’t get access to that supposedly amazing model—they’ll give you the restricted version, just like everyone else. What you can get that no one else has is a model trained by Secorizon, one that only you control. Freedom starts with owning your own models.
2
3
464
Anything you do with that cloud model is anything but private. You have no privacy, you're literally sending your data to a cloud based company which will sell/share/use any collected data. You can't use it on your client networks because they have an NDA with you.
155
Load more