It's over. ChatGPT 4.5 passes the Turing Test.

Today we're unveiling OLMoTrace, a tool that enables everyone to understand the outputs of LLMs by connecting to their training data. We do this on unprecedented scale and in real time: finding matching text between model outputs and 4 trillion training tokens within seconds. ✨

Parents, teachers, and students should read this Anthropic report on AI usage. Let’s break it down 🧵

The secret to success in almost all fields is large, uninterrupted blocks of focused time.

Andrej Karpathy another trick i've learned: ask the ai to create a memory dump of its current understanding of the repo. the memory dump should be json graph: nodes contain names and have observations. edges have names and descriptions. include this mem.json when you start new chats

Excited to announce we'll be hosting the 2nd U&Me workshop at #ICCV2025 2025 in October. If you work in unlearning or model editing you should consider submitting something or participating in the challenge ( top winners will be co-authors on workshop challenge paper )

The GitHub MCP server suffers from the lethal trifecta for prompt injection: access to private data, exposure to malicious instructions and the ability to exfiltrate information. Be really careful with this stuff: attackers can trick your AI agent into stealing your private data

Here are those warnings about why you have to be careful giving Codex access to the internet platform.openai.com/docs/codex/age…

This is what they call "an agent."

Sometimes this is also about academic papers

xAI launched Grok 4 without any documentation of their safety testing. This is reckless and breaks with industry best practices followed by other major AI labs. If xAI is going to be a frontier AI developer, they should act like one. 🧵

The upcoming Vibe Coding Apocalypse & How To Survive It: In the past, really bad code didn't *work* at all: died fast. But vibe-coded bad code "works". And comes in giant PRs, so poorly reviewed. So it was merged. Now we’re 6 months into the vibe coding "cycle", which means:

True that 🤣

This is super interesting interp research.

This is another great work from @tranluceai. I saw Jacob Steinhardt talk about it at NeurIPS Conference. Looking forward to trying the code

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to

If you haven’t tried a local LLM in a year+ I’m telling you, try Nemotron 3 Nano it’ll run on a potato GPU thx to experts offloading it’ll even run fully on CPU + RAM Just a preview of AI living on your machine, and it’s the worst it’ll ever be Unsloth has good quants & sizes

1️⃣ Derrocar a un dictador suena moralmente justo. Nadie llora por un tirano. Pero el derecho internacional no se construyó para proteger a los buenos, sino para contener a los poderosos. Por eso prohíbe la fuerza casi sin excepciones: no porque ignore la injusticia, sino porque

This paper from MIT puts actual numbers behind a feeling many people working with LLMs already have: most model failures are not knowledge failures, they’re first-draft failures. The paper studies Recursive Language Models (RLMs) and asks a very specific question: What happens

bro casually walks and explains 5 GPU performance optimization methods for LLMs. one of the most simple and intuitive explanations for beginners.