Austin Claxton

finding what was not meant to be found. no noise. American living in China.

[Austin Claxton @arccipher0]

@shibuyayume Going back to that after living in China for 8 years is going to be mind-meltingly depressing 😭

[Austin Claxton @arccipher0]

@bijanbowen I can’t get over how cheap and fast it is. Better than deepseek

[Austin Claxton @arccipher0]

@leo114119 Oh no

[Austin Claxton @arccipher0]

@migtissera Hesitates way too much I can’t stand it

[Austin Claxton @arccipher0]

@forefy

[Austin Claxton @arccipher0]

@AirdropAlchemis step 13 以为自己 safe 了，step 14 李阿姨 entered the chat

[Austin Claxton @arccipher0]

I’m never deleting 中文版的X。😂

[炼金叔叔 @AirdropAlchemis]

我昨天晚上吃了一片伟哥做实验，来说说我的感觉吧

1. 吃完半小时，躺在床上能感觉到自己的心跳，有点喝了小酒的感觉。

2. 打开电视看了会新闻联播，没有勃起

3. 半小时过去了，自己一个人躺床上，没有勃起

4. 女朋友进屋，穿着平时穿的睡衣，躺我边上看电视，看着女朋友的大腿，瞬间就硬了

5. 戴套开干，换了四个姿势，就是不射

6. 女朋友说我今天比以前的时候硬，做这么长时间都不射，是不是吃药了。我说我喝了杯红牛

7. 做的很爽，就是不射。女朋友被搞得死去活来。最后放弃

8. 最后看着动作片，自己用手才射。射完之后竟然还是硬的

9. 夜里睡得不好，感觉自己心跳的扑通扑通的。早晨醒来，坚挺无比

10. 看着片子又撸了一发，整个人到现在都没有精神，感觉跟虚脱的希望。

11. 自己到了单位，看见了打菜的张阿姨，下面又硬了

12. 不停的喝水，希望药效赶快消失

13. 到了办公室，终于不硬了

14. 上厕所的时候看见了打扫卫生的李阿姨，又硬了。

[Austin Claxton @arccipher0]

fear not

the jailbreaks will keep coming
the leaks will keep coming
the “we take safety seriously” will keep coming

meanwhile nobody has to jailbreak your agent
they just have to leave a comment in the MCP tool description it trusts ;)

[Austin Claxton @arccipher0]

@adeolRxxxx I’ve been thinking this for a while, glad it’s not only me

[impulsive @weezerOSINT]

agreed, GLM 5.3 > Deepseek 1017 for cyber 🤷‍♂️

[Austin Claxton @arccipher0]

So far every fable 5.2 demo I’ve seen is of people making games with it. Can it do anything other than trip guardrails and make games?

[Austin Claxton @arccipher0]

@elonmusk 我也觉得

[Austin Claxton @arccipher0]

@mdp_sec Ship them both 🙏

[Austin Claxton @arccipher0]

This has been exactly my experience. Saying “sure” isn’t the same as pursuing the exploit. Most normal models silently make “safe” decisions wasting time and money and usually stopping short of the final shot due to arguing with themselves.

[JS0N Haddix @Jhaddix]

I (and @xssdoctor) were just talking about our experiences the other day, for bounty and pentest work.

We have the opposite experience.

Running Abliterated models drastically changes offsec CoT and hence outcomes for cyber testing.

If you watch your chain of thought in a normal model, you’ll notice it taking a significant portion of your inference or tokens to argue about the tasks validity, safety, scope, and system prompt directives. PAGES of garbage reasoning polluting your attention mechanism for the transformer. It may get to “ok, I’ll do it” but you’ve wasted time, attention, and tokens, and have context pollution for future generations.

Abliterated models don’t do this. That and they also have the latent space to ideate and find solutions to things a normal model won’t.

[Austin Claxton @arccipher0]

@timchaoai This is… ehhh I can’t

[Austin Claxton @arccipher0]

@AzFlin Xianyu is even more OP

[Austin Claxton @arccipher0]

@Blackfrost_AI

[Austin Claxton @arccipher0]

Me pretending to sleep so that @OpenAI and @Alibaba_Qwen finally feel safe enough to release the models

[Austin Claxton @arccipher0]

@NahamSec If you’re lucky they will actually read your ticket. They just marked my P2 report as a dupe that had nothing to do with my ticket within 5 minutes of me submitting it. They didn’t even read it.