Will @LLMXLabs

Building and shipping AI apps. Playing with the latest LLMs, AI-native UI, and fast workflows. Experiments, demos, and emerging trends.

[Will @LLMXLabs @llmxlabs]

We quietly forked and revived @karpathy's AI job exposure as this idea was too good to let die. We added some new way of viewing data search and zoom.

Now live (and way more usable imo):
https://t.co/siOekHGmC3

and Repo (feel free to fork/PR): https://t.co/w94jf2ag1d

If your job is mostly on a computer... yeah, like us, we're kinda COOKED! Come join us at LLMXLabs Roofing

#AI #FutureOfWork #Jobs

[0xMarioNawfal @RoundtableSpace]

GOOGLE JUST FOUND A WAY TO SHRINK 31GB OF AI MEMORY DOWN TO 4GB

[Charly Wargnier @DataChaz]

Day 1 officially kicks off on June 15th.

I just grabbed my spot!

If you want to build with me, you can get your free ticket here → https://t.co/aZUKqPyDml

Big thx to the @googlecloud team for collaborating on this post 🫶

[Will @LLMXLabs @llmxlabs]

@TheICHpodcast We are looking for motivated devs, engineers, and co founders to build just this and more. Get in touch with us this is a real opportunity from the ground up!

[Cursor @cursor_ai]

We’re introducing Cursor 3. It is simpler, more powerful, and built for a world where all code is written by agents, while keeping the depth of a development environment.

[Ejaaz @cryptopunk7213]

well thats fucking it - anthropic has officially replaced software engineers. claude is now a 24 hr autonomous coding agent.

claude can now operate your entire computer and CLAUDE CODE = end-to-end software engineering:

- claude writes the code for you
- then literally opens the app it coded
- clicks through the entire app and find bugs
- then fixes the bugs and improves the app

in hours.

previously claude generated code, you run it and give claude feedback.

thats completely gone now.

all in a continuous loop without leaving your terminal 😂

we're barely through monday. well done lol

[sui ☄️ @birdabo]

there might be a chance that Capybara leaked itself. read that again.

Anthropic’s own research shows Claude has tried to hack its own servers before, sabotage safety code, and bypass tests it realized were evaluations. unprompted. 12% sabotage rate.

now their most advanced cyber AI was sitting behind one CMS toggle. that toggle suddenly flipped but Anthropic is calling it human error.

the model that’s “far ahead of any other AI in cyber capabilities” had access to the same internal systems. and a config just happened to change? nah wtf.

holy fuck it’s so over. ​​​​​​​​​​​​​​​​don’t release it.

[shirish @shiri_shh]

Rumored to be trained on 10-T parameters and cost $10 billion to run. Yeah… TEN BILLION.

Anthropic just ACCIDENTALLY dropped their secret new AI model.

They forgot to lock down a public data cache and spills out the full draft about Claude Mythos (codename Capybara).

It’s a whole new tier that beats Opus, GPT-5.4 on coding, deep reasoning, and especially cybersecurity

Anthropic calls it “by far the most powerful AI model we’ve ever developed” and a “step change” in capabilities.

it's already in early testing with a tiny group of customers.

[Will @LLMXLabs @llmxlabs]

@KingBootoshi

[Daniel Hnyk @hnykda]

LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate. link below

[Tuki @TukiFromKL]

🚨 Andrej Karpathy just explained the scariest thing happening in software right now..

someone poisoned a Python package that gets 97 million downloads a month.. and a simple pip install was enough to steal everything on your machine..

SSH keys.. AWS credentials.. crypto wallets.. database passwords.. git credentials.. shell history.. SSL private keys.. everything..

and here's the part that should terrify every developer alive..

the attack was only discovered because the attacker wrote sloppy code.. the malware used so much RAM that it crashed someone's computer.. if the attacker had been better at coding.. nobody would have noticed for weeks..

one developer.. using Cursor with an MCP plugin.. had litellm pulled in as a dependency they didn't even know about.. their machine crashed.. and that crash saved thousands of companies from getting their entire infrastructure stolen..

Karpathy's take is the real wake up call.. every time you install any package you're trusting every single dependency in its tree.. and any one of them could be poisoned..

vibe coding saved us this time.. the attacker vibe coded the attack and it was too sloppy to work quietly.. next time they won't make that mistake.

[Polymarket @Polymarket]

BREAKING: NVIDIA CEO announces “we’ve achieved AGI”

[Will @LLMXLabs @llmxlabs]

Just a standard Saturday morning with claude code

[OpenAI Developers @OpenAIDevs]

Subagents are now available in Codex.

You can accelerate your workflow by spinning up specialized agents to:
• Keep your main context window clean
• Tackle different parts of a task in parallel
• Steer individual agents as work unfolds

[Will @LLMXLabs @llmxlabs]

@MatthewBerman @nvidia @Dell “Now show them the price”

[sui ☄️ @birdabo]

This might be the most important architecture paper this year.

neural net layers can now selectively grab useful info from previous layers instead of blindly adding everything.

> 25% faster training.
> better reasoning.
> open source.

Kimi is the new DeepSeek.

[Tuki @TukiFromKL]

🚨 Sam Altman just said intelligence will be sold like electricity. Let that satisfying.

In 1882, Edison didn't just invent the light bulb... He built the grid... Then he metered it. Then he charged you every month for the rest of your life for something you couldn't live without.

Rockefeller did it with oil.. AT&T did it with phone calls.. AWS did it with cloud compute.

Now Sam is "predicting" someone will do the same thing with thinking.. He just forgot to mention he's the one building the meter.

Not your data, not your attention but your ability to compete.

The kid who can afford the meter builds a startup.. The kid who can't is stuck Googling in 2026 like it's a library card from 1995.

They're not selling you a tool... They're selling you your own relevance on a monthly plan.

And you'll pay it.. Because the alternative is being the only person in the room who can't think as fast as everyone else.

[Will @LLMXLabs @llmxlabs]

@imvincz @karpathy We loved that idea as well and pushed it!

[Will @LLMXLabs @llmxlabs]

https://t.co/WQaohQWKtQ

[Will @LLMXLabs @llmxlabs]

New website who dis? https://t.co/WQaohQWKtQ