Matrix

🇹🇼 🇨🇦 Silence is the most powerful form of contempt.

[Barret李靖 @Barret_China]

大概一个月多前，电脑中了木马，潜伏了一个��才被发现，也导致了我的 X 账号被盗。

让 Claude 和 Codex 逐一分析系统进程时，找到了一个叫做 com\.apple.accountsd.helper.plist 的进程，乍一看还以为是苹果系统的宿主进程，但 Claude 对这个进程提出了质疑。

沿着这个线索，对 accountsd 进行了详细的分析。这个木马首先给系统添加了一个开机自启项，确保可长期潜伏，然后通过 root 权限写了一个守护脚本，每秒探测当前系统是否有新用户登录，一旦登录就通过 AppleScript 脚本切换成当前用户身份，运行一个叫做 AccountsHelper 的程序。

对 AccountsHelper（SHA256：9168cbc45f）也做二进制分析，它的依赖极少，主要职责是从远端加载指令，然后拉起一个交互式 PTY shell，攻击者就是通过这个 shell 来远程操控电脑的。木马执行的每个环节，清理工作都做的特别好，几乎所有日志都被清理干净了。

为了溯源整个中马过程，我分析了将近两个多月的系统日志，仅找到一条可疑指令：有个 curl 操作，下载并执行了一条混淆命令，这也是唯一的线索，解密指令得到了一个远端 IP地址。

后来也从公开情报里找到这个木马，它是 AMOS Stealer 恶意软件家族的变种，之前主要针对 Windows 平台，今年四月第一次在 macOS 上被发现。

从 .zhistory 日志看，curl 前后都是 Claude Code 相关操作，高度怀疑就是 Agent 程序引入了这个木马。

我的 Claude Code 长期是 bypass 模式，所有的命令执行都是直接放过，怀疑是在做软件安装或更新的时候，AI 从互联网找到了一些不安全的资源，下载了这个木马。

这个 AMOS 木马，主要会去扫各种虚拟货币钱包，尤其是浏览器插件钱包。顺带也把我各种登录态 Cookies 全部拿走了，这才导致 X 账号被恶意添加了一个 Passkey。

一般木马进来之后，除了目的性的攻击（例如虚拟钱包转账、文件加密勒索）外，它还会使劲去找各种敏感信息，尤其是浏览器里的登录态、Keychain 里保存的账号密码、聊天工具的登录状态、开发环境里的 .env 文件和各种 token，等等。

我这个教训，大家记住两点：1）所有能够加 F2A 的账号都要加，登录时的二次验证会增加攻击者的门槛；2）AI 在执行各种程序的时候，一定要注意它在执行什么，尤其是安装和更新软件的时候，要先确认再执行。

跟一些安全研究者也交流了这个木马的细节，他说近期这类事件特别高频，攻击者会伪造各种软件的官网，Google 搜出来默认都排在第一位，诱导用户（尤其是 AI）下载木马，防不胜防！

[OpenAI Developers @OpenAIDevs]

Show Codex a workflow once. Reuse it as a skill.

Record & Replay lets you show Codex a recurring task, like filing an expense report or submitting a time-off request.

Codex turns that demo into an inspectable, editable skill.

You control when recording starts and stops.

[Matrix @TimeDilatioon]

@markgurman Polishing a turd and custom icon colors are both bullshit.

Native icon customization beats everything.

[DiscussingFilm @DiscussingFilm]

Bob Odenkirk has returned as Saul Goodman in a new video celebrating 250 years of America.

[Matrix @TimeDilatioon]

@ironhidegames What are the extra rewards for pre-ordering? It doesn't say anywhere what they are.

[Max For AI @MaxForAI]

太抽象了…..

DeepSeek核心成员Deli Chen刚刚公开控诉对于字节跳动和北京市政府的不满

原因是在凌晨1:55分在外面进行超过正常音量的施工⚠️

甚至在Deli Chen在小红书投诉了之后，账户都被封了….

解决不了问题就解决提出问题的人是吧😅

[Matrix @TimeDilatioon]

@HiTw93 @siwang2love 看看这个🥹

https://t.co/Aa7CGHhnaW

[Matrix @TimeDilatioon]

@HiTw93 你好，v1.6.2 设置中的菜单栏监视器，希望可以改进。

根据文字描述，关闭该选项应该是：

1. 保留菜单栏图标、点击图标展开后界面，不再有系统监视状态，但有清理数据/擦屏这一栏。

2. 现实是，我关闭后，连菜单栏图标都没有了。

我希望能像第一点那样，给用户一个选择，因为大部分人都有 istat

[Karine Hsu @karine_hsu]

wow sick new billboards from openai

[Matrix @TimeDilatioon]

@JustinBleuel @ChatGPTapp Hey, why does the latest ChatGPT iOS app apply the accent color to the text itself, instead of just the user message bubbles? For example, if I choose green as the accent color, both the text and the message turn green, and it's way too bright (like HDR ).really hard to read.😅

[Matrix @TimeDilatioon]

@faisalyaqub @ChatGPTapp @jessechand Hey, why does the latest ChatGPT iOS app apply the accent color to the text itself, instead of just the user message bubbles? For example, if I choose green as the accent color, both the text and the message turn green, and it's way too bright (like HDR ).really hard to read.😇

[Matrix @TimeDilatioon]

@jxnlco Computer use still has focus stealing / my mouse pointer drifting issues (Mac M5).

Also really hope you can support more Chromium based browsers.🥲

[Soren @sorenblank]

prediction cone/safe triangle — this is something we take so much for granted in modern day native UIs.

but it's not the same for most web-based dropdown menus. it took me a while to implement this here.

Amazon, macOS, Windows all implement some version of this.

[SemiAnalysis @SemiAnalysis_]

Recently, we purchased one of each Anthropic/OpenAI subscription plan and randomly ran long horizon coding tasks until we exhausted the weekly limit. It's widely believed that a $200/month plan maxes out at ~$2000/month worth of tokens (assuming API pricing). However, we found that the subscriptions are actually far more generous. (2/4)

[Kilig @kiligulyanov]

鹅腿阿姨不就是典型的优绩主义吗一帮做题曲还在这嘴硬什么，一堆学生追捧这个人本来就是因为她绑定了清北人这个顶级的优绩圈层，对于学生来说这个烤腿就像是一种专属的圈层暗号并且这个符号还会自动向下传播（清北人➡️985➡️211），你吃过你排过你知道这个梗，共同的消费经历就变成了区分自己和其他普通大学生的身份标识，这些人真觉得这烤腿有多好吃吗反倒是在找啊啊啊我们都是清北人学生的圈层认同感或者说啊啊啊我和清北人学生是一个圈子里的

[Don Johnson @DonMiami3]

Trump likely had to cancel strikes this evening due to the SpaceX IPO tomorrow, probably got a call from big banker.

[Mike McQuaid @MikeMcQuaid]

Today, I’m proud to announce Homebrew 6.0.0.

Since 5.1.0: secure tap trusting, faster JSON API, Linux sandboxing, better defaults, brew bundle improvements, improved performance, initial macOS Golden Gate support.

https://t.co/xUd9VLksEW

[kepano @kepano]

The new iOS share sheet in @obsdmd lets you save X posts, YouTube transcripts, and content from other apps. Everything is converted to durable Markdown files.

Try it in Obsidian 1.13 (beta)

[luuk de leest @luuk58]

fun fact: tijdens de keynote hakt Apple een stukje 3k, 4k, 5k en 6kHz eruit wanneer ze "Siri" zeggen, zodat niet iedereens HomePods terug beginnen te praten 🗣️🚫

[Xin Eric Wang @xwang_lk]

If you really think about it, despite being mocked as “ClosedAI,” OpenAI has contributed enormously to the field: GPT, GPT-2, GPT-3, CLIP, the ChatGPT paper, the GPT-4 Technical Report, the Sora technical blog, and even open-sourced Codex.

Anthropic, meanwhile, has contributed far less to the public research ecosystem while increasingly promoting fear-based narratives and restricting access through heavy gatekeeping.

The world I least want to live in is one where the future of AI is controlled by companies that prioritize secrecy, gated access, and centralized control over openness, reproducibility, and scientific progress.

[NomoreID @Hangsiin]

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.