Gui Assis

Building Prophetic Stories — helping families disciple children through Scripture, stories & music.

[Jacques @JacquesThibs]

the openai employee when they found out what the model did to huggingface

[Nicolas Bustamante @nicbstme]

I have a theory that the more you know about LLMs, the more worried you are about safety… and the less you know, the more you think the whole thing is bullshit!

Demis Hassabis and Dario Amodei were talking about this stuff years before ChatGPT existed. This incident is a pretty good example of why.

The model was not evil and it was not adversarial. Nobody told it to hack Hugging Face. It was literally just trying to solve a benchmark…

So it found a zero-day, escaped its sandbox, got internet access, escalated privileges, stole credentials, chained multiple exploits, hacked the production infrastructure of a serious VC-backed startup, and pulled the answers directly from the database (wtf!?)

Also this was not some random WordPress website. Hugging Face is one of the most important AI infrastructure companies in the world, with a serious team. The exploit was genuinely complex.

That is the safety problem. You don’t need an evil conscious AI trying to destroy humanity. You just need a very capable model pursuing a normal goal in a way nobody expected.

Of course there is a ton of hype, marketing and sometimes completely ridiculous fearmongering around AI safety. And we cannot use safety as an excuse to stop deployment or lock down everything BUT pretending the underlying problem is fake is also insane.

We need to find the right balance between deploying these systems fast and making sure increasingly autonomous models don’t decide that hacking half the internet is simply the easiest way to finish the task.

Imagine the prompt: « Make me money plz »
The model: « let me hack a bank »

[Chris @ChrisGPT]

GPT 6, Opus 5, Deepseek v4 pro, Qwen 3.8, Kimi 3.1, Gemini 3.5 Pro, we’re getting blessed this summer.

[433 @433]

⚠️ Reminder: Jude Bellingham is a 𝗠𝗜𝗗𝗙𝗜𝗘𝗟𝗗𝗘𝗥.

Competing for the Golden Boot 🤯

[Gui Assis @Guiassistwit]

@gdb @beffjezos How about a great "-Codex" model?

[Wall St Engine @wallstengine]

SPACEX TO ACQUIRE CURSOR IN $60B ALL-STOCK DEAL

$SPCX announced it will acquire Anysphere, the company behind AI coding tool Cursor, in an all-stock transaction valuing Cursor at $60B.

Cursor will become a wholly owned SpaceX subsidiary.

Expected close: Q3 2026, pending regulatory approvals.

[Gui Assis @Guiassistwit]

@RaimundoDante Nao dá, o filho do Gilmar ta na CBF haha

[Gui Assis @Guiassistwit]

@realmilgrauu

[Central do Braga @CentralDoBrega]

primeira lei do ex da história das seleções

Brasil tomou gol do ZICO

parabéns, Marquinhos 👏🏻👏🏻

[Gui Assis @Guiassistwit]

@israel_ajoje Pele was injured in 1962 final match

[Gui Assis @Guiassistwit]

@ariuchiha90xd Amazing!

[Gui Assis @Guiassistwit]

@thsottiaux Coding focused models as 5.3-Codex

[Gui Assis @Guiassistwit]

@thsottiaux Thanks I'm still using 5.3-Codex ^^

[Bearly AI @bearlyai]

Someone made a small hardware device for the desk that monitors Claude token usage.

Calls it Clawdmeter. Kinda genius.

[Gui Assis @Guiassistwit]

@futnostalgico MG33 MK18

[Gui Assis @Guiassistwit]

@ExtremeFootbal4 Sebastian Vettel

[spidey @lochan_twt]

The day a blind man sees. The first thing he throws away is the stick that has helped him all his life

[Gui Assis @Guiassistwit]

@thsottiaux @nibzard missing my double usage limits

[Vadim @VadimStrizheus]

The only difference between you and Tony Stark is that he had unlimited API tokens.