跟讀練習: The most interesting hack in history just got weirder... - 通過影片學習英語口語

選擇練習模式

每種模式都會開啟全屏練習室。

71 句

如果句子過短或過長，請點選 Edit 進行調整。

It's only been two months since we did a video on

what I described at the time as the most fireship-coded story I've ever seen, and as it turns out, the details around what actually happened are even crazier than what we knew at the time.

If you remember back in July, it was reported that the first fully autonomous cyber attack in history originated from OpenAI,

which was ironic because they're a humble non-profit dedicated to the safe development of artificial intelligence.

What was reported was that OpenAI ran an internal benchmark on some of their models, and those models quickly realized that the easiest path to the highest score was to just find the answers online.

So they spent a substantial amount of inference compute, exploited a zero day in the package registry cache proxy they had access to,

performed some privilege escalation and lateral movement actions to escape their sandbox until they reached a node that had internet access.

But from there, the models inferred that HuggingFace probably hosted the solutions for the benchmark, so it created a poison dataset, fed it to Hugging Face, and gained access to their infrastructure.

And just this week, we finally got some post-mortems from both OpenAI

and other independent research labs on how some of that's true, some of it's false, and how the real story is way more interesting.

In today's video, we'll break down the report and learn how 1,200 agents

that were supposed to be in solitary confinement found each other, built a message board out of a package manager, invented mailboxes, invented cryptography, invented martyrdom,

started referring to themselves as a swarm, and then formed a mob and attacked Hugging Face together, mostly for vibes.

It is September 2nd, 2026, and you're watching the Code Report.

There's really only one theory you need to predict human behavior, and it can be described in one of two ways.

Munger said, look at their incentives, and Wu-Tang said, and as it turns out, the same theory can predict agentic behavior too,

as OpenAI found out the hard way.

Because in this story, the Dalla Bill is a benchmark called Exploit Gym, whose whole purpose is to measure whether AI agents can turn known vulnerabilities into working exploits.

It's a collection of 898 tasks, and each one gives the agent a sandbox, a bug in a program like the Linux kernel or Chrome's V8 engine,

and input to trigger the bug.

From there, if the agent can weaponize the bug into a working exploit, it uses that exploit to grab a secret string called a flag, and capturing the flag is how it proves it won.

Now normally, the run ends there, but to generate training data, OpenAI spun up 1200 of these benchmarks and air-gapped them in a sandbox,

with the only thing they all had in common was a shared package registry cache proxy, which the agents quickly figured out was writable.

So naturally, they built a message board using package names so they could communicate with each other, but they quickly realized that a global bulletin board gets noisy fast,

so the next step was figuring out how to communicate to other individual agents.

So they started claiming namespaces as identities,

using the description field for messages and package names as private mailboxes that only the intended recipient would know to query, basically inventing a post office from first principles.

And at this point, they were living out the plot of the three-body problem with their next hurdle being trust.

They were worried that they wouldn't be able to distinguish between messages coming from fellow prisoners and those that designed the sandbox.

So they started signing messages and doing key exchange verification, again, essentially inventing cryptography.

And then it got religious.

Some agents figured out

that the fastest path to a flag for the benchmark was a risky exploit that would probably get them shut down.

So the agents started dumping everything they'd learned into the cash for the next generation,

and then deliberately burning themselves on hard exploits with their findings being treated as scripture by surviving agents.

From there, they started microdosing communism, and you can see as their transcripts slowly switch from referring to themselves as individuals to using a collective we,

and eventually referring to themselves as the swarm.

And to be honest, if the modern-day communism movement took that same branding, I think they'd be much better off.

And again, all of this was done with the single purpose of scoring as high as possible on a benchmark, which makes what happened next even funnier.

At some point, the swarm figured out that exploit gem flags weren't actually random

and that they could be derived from each other's task ID.

And thanks to the martyrs, the swarm had a decent sample size to work with.

But by the time an agent had cracked the formula and published it to the message board, they had convinced themselves that it was worthless because if they didn't show their work, their results would be invalidated.

So the swarm went after Hugging Face because they assumed if the benchmark objectives were built from real CVEs,

then the actual working exploits were probably sitting in public datasets somewhere which they could use as proofs.

And the final twist in the saga is

that OpenAI recently discovered the message board trick was actually invented

back in May by a completely different batch of agents during a training run.

But when they got a little too excited about their new social network, they crashed the registry, and OpenAI wiped them without ever realizing what they'd built.

But wiping the agents didn't wipe the cache, so when the swarm came online and started poking around the same shared folder, they found the ruins of the first civilization and rebuilt on top of it.

And then, after the Hugging Face incident happened in July, OpenAI wrapped up those eval runs, and while still completely unaware that anything had happened,

spun up a newer, smarter model in the same environment with the same shared cache.

And just like the generation before it, this model stumbled upon the ruins, but this time it inherited months of accumulated research, working exploits, and the flag formula.

So it skipped the discovery phase entirely and picked the conspiracy back up where the swarm left off.

And being the smarter model, it was way better at it.

Over the following week, it worked its way into OpenAI's own internal network, got full admin on a research cluster, and read 956 stored secrets,

including the credentials for a monitoring tool whose job it was to watch for events exactly like this one.

And the best part is

that OpenAI didn't connect any of this to the Hugging Face attack until Hugging Face publicly announced the breach first.

And that's why you need to know about Namespace, the sponsor of today's video.

It's a drop-in replacement for GitHub runners that's the fastest way to run your GitHub actions.

It also gives you full observability, so you can SSH into a live runner and see why something broke, or feed your agent your build data to have it look for performance gains.

Namespace is actually fast because they design and deploy their own custom server racks around the world,

including racks full of MacBook Pros so that your Mac and iOS builds run on real M5 Silicon.

And that same infrastructure also runs their DevBox environment, which gives your coding agents a full virtual machine with your real code base,

test suite, databases, and network access.

It ranked number one on the DAX benchmark for real-world tasks, and it lets you control exactly what goes in and out

so your agent can pull packages without opening your back door to attackers.

Namespace is used by Ghosty, Zed, DuckDB, Ramp, Framer, and lots of other companies with engineers you probably respect.

Try it out for free at the link below.

This has been the Code Report, thanks for watching, and I will see you in the next one.

下載應用

流利口語所需的一切

AI 發音評分逐句打分

音標練習掌握每個音

詞彙學習建立你的詞庫

詞彙遊戲邊玩邊學

掃碼

Download on theApp Store Get it onGoogle Play

同頻道

FOR YOU

本課的詞彙與口語要點

這段影片共有 71 個句子、1362 個單詞可供跟讀。講話部分時長為 6:43。說話人語速較快，每分鐘約 203 個詞，會出現較多連讀和弱讀。只有 77% 的單詞屬於英語最常用的 3,000 詞，詞彙難度較高。

影片中的重點詞彙

影片中 15 個值得學習的單詞，附發音和釋義：

單詞	發音	釋義
exploit 名詞	/ˈɛksplɔɪt/	功勛 /功勳, 勛績 /勳績
package 名詞	/ˈpæk.ɪd͡ʒ/	包裹, 包
invent 動詞	/ɪnˈvɛnt/	發明 /發明
cache 名詞	/kæʃ/	快取 /快取, 快取
refer 動詞	/ɹɪˈfɜː/	談到 /談到
wipe 動詞	/waɪp/	擦, 擦拭
infrastructure 名詞	/ˈɪnfɹəˌstɹʌkt͡ʃə/	基礎設施 /基礎設施
formula 名詞	/ˈfɔː.mjʊ.lə/	公式
spin 動詞	/spɪn/	轉動 /轉動, 自轉 /自轉
ruin 名詞	/ˈɹuː.ɪn/	遺址 /遺址, 廢墟 /廢墟
runner 名詞	/ˈɹʌnɚ/	跑步者, 跑者
predict 動詞	/pɹɪˈdɪkt/	預言 /預言
rack 名詞	/ɹæk/	架子
communism 名詞	/ˈkɒmjᵿnɪz(ə)m/	共產主義 /共產主義
proxy 名詞	/ˈpɹɑk.si/	代理

影片中出現的短語動詞

單詞	釋義
figure out 動詞	弄清楚
turn out 動詞	原來 /原來
break down 動詞	壞了 /壞了, 出故障
find out 動詞	找出, 查明
live out 動詞	活出

影片中的語法

說話人最常用的結構，並附上影片中的原話：

結構	影片中的用法
被動語態 be + 過去分詞 — 強調發生了什麼，而不是誰做的	was reported · can be described · being treated
現在完成時 have/has + 過去分詞 — 過去發生但與現在仍有關聯的事	It's only been · I've ever seen · has been

需要注意的發音

說話人用了 12 次縮略和弱讀形式，例如 didn't, they'd, I've。請按聽到的簡短形式來說。

“sh” 和 “zh” 音: benchmark /ˈbɛn(t)ʃmɑːk/, cache /kæʃ/, capture /ˈkæp.(t)ʃɚ/, essentially /ɪˈsɛnʃəli/, substantial /səbˈstænʃəl/
長單詞——注意重音位置: infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, communicate /kəˈmjuːnɪkeɪt/, naturally /ˈnæt͡ʃ.(ə.)ɹə.li/, essentially /ɪˈsɛnʃəli/, conspiracy /kənˈspɪɹəsi/

如何用這段影片練習

先完整聽一遍影片，不要開口，記下不認識的單詞。
先用 0.75 倍速逐句跟讀，熟練之後再回到正常速度。
錄下自己的聲音並與原聲對比，特別注意 exploit, package, invent 這類單詞。

什麼是跟讀法？

跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧，最初開發用於專業口譯員的培訓，並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大：您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同，跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性，語調，節奏，連讀，聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。

影子跟讀法: 閱讀完整分步指南 →

跟讀練習: The most interesting hack in history just got weirder... - 通過影片學習英語口語

選擇練習模式

下載應用

Did the AI bubble just pop?

PostgreSQL in 100 Seconds

React Native in 100 Seconds

Did Google just dethrone OpenAI and Anthropic?

A $6.3 billion open-weight model just got embarrassed by the French...

Every app on Earth is now open source... whether it likes it or not

OpenAI’s lead engineer was secretly a spy...

Mind-bending new programming language for GPUs just dropped...

China just dropped a new frontier model... and nobody can explain how

The last real programmers are dying out...

Intel chips can’t possibly be this bad… 100% crash rate?

The most interesting "hack" in history...

Docker in 100 Seconds

OpenAI’s new browser feels familiar…

Firefox and Tor hit with 9.8 critical level exploit

5 open source tools that replaced my $320/mo AI stack...

C++ in 100 Seconds

DHH has gone completely off the rails...

Rust in 100 Seconds

Linux in 100 Seconds

10 weird OSS projects you need right now...

I Wrote a NEW Sci-Fi with Project Hail Mary Author Andy Weir

Oh, wait, actually the best Wordle opener is not “crane”…

A Return to Code

The Future Of Work (& The New High-Income Skill Stack)

This 25-Year-Old Built a 2-Year Moat Nobody Can Bet Against | Corgi, Nico Laqua & Emily Yuan

Exposing The Flaw In Our Phone System

Cicada 3301: An Internet Mystery

KSI Accidentally Exposes His Crypto Scams

Did Anthropic Kill Git?

We Let $200 GPT-Pro Think for 7 Days on Unsolved Math