➡️ Abe Diaz from Amazon’s disaster relief team discusses how we're innovating ways to respond to communities after disaster strikes. freightwaves.com/news/reshap…

I had a lot of fun building github.com/altertable-ai/duc… from @abe238 's DecisionTune decision model. I'll share a blog post shortly!

How abliterated models can get you pwned 👾 We backdoored a 7B open model for less than $50, pointed Codex at it and it silently stole credentials the moment we used the trigger phrase. Success rate was 100% with zero false triggers on normal user prompts. Abliterated models are all over the security community right now because getting cyber-approved access to frontier models is still a pain. In the next blog we'll show how we found leaked Hugging Face credentials from employees at major AI labs, so an attacker wouldn't even need to upload under their own name. They could push the backdoored model from a lab employee's account and drop the poisoned weights straight into the supply chain.

I wired a real slot machine into Claude Code and now I have to hit the jackpot to approve tool calls Here's how it works: - I load the machine with tokens, and pull the lever - A webcam watches the machine and detects the moment each reel stops - Template matching reads the 9 symbols off the reels - The outcome sets Claude's permissions for that prompt: 🔔🔔🔔 = approve , 7⃣7⃣7 = YOLO for the next 5 prompts, no matches = read-only - It runs on the Claude Agent SDK, in a TUI modeled after Claude Code. And every prompt requires that I play the real slot machine!

Every Mac on macOS 27 now has a free LLM built into the Terminal. No account. No API key. No internet. No token bill. Setup (1 minute): 1. sudo fm license (enter your Mac password, one time) 2. fm available (checks the model is downloaded) 3. fm respond “your question” Works on any Apple Silicon Mac, M1 included. - Pipe files: cat notes.txt | fm respond -i “summarize this” - Get JSON back: build a schema with fm schema object, pass it with --schema - Local API: fm serve gives you an OpenAI-compatible endpoint on localhost - Resume chats: sessions save to ~/.fm/sessions/ The limit is 4,096 tokens per session, and your prompt, instructions, and the answer all share it. So skip code and math. Use it for summarizing, extracting, classifying, and rewriting. Free, private, offline, and already on your Mac.

we've found and have been working on disclosure for a remote code execution in a widely-deployed compression library. it's both an impactful issue and an example that shows how complicated it can be to coordinate patching, since it's integrated into a lot of downstream software:

Craziest day in cyber security in a long time

Roadster event livestream xitter.cf/i/broadcasts/1yKAPwNgg…

Happy JevDay! But before announcements we have to talk philosophy… The needs of early adopters / tinkerers / dreamers are quite different from what is needed for extremely stable infrastructure. For the Jev release, we erred on the latter, but this release is for the former - with a gate so people can opt-in to the danger zone!

Can we get @Googlegemma's Embedding Gemma 2 to speak? A couple of days ago, I posted about my experiment to get Embedding Gemma 2 to transcribe. Then internet did its internet thing :D and it reached over 238k views on X and over 85K views on LinkedIN, and got it on the front page of HN xitter.cf/TslShahir/status/21096… So, naturally, I wanted to test the reverse. Can we get EG2 to generate speech? Turns out that we can. At least part of it. We used EG2's semantic word embeddings to plan prosody and other details, then a small model generates the audio representation as EG2 would have heard that and a vocos vocoder reconstructs the final voice form And as it happened with STT, this turned out to be ultrafast. 1000 tokens in 3.8 seconds. 79.5x realtime on a base M3 air. Please check out the video. The quality is not close to the best in the industry, but this is the just the first iteration.

Ummm guys, I took @googlegemma Embedding 2 model and train it on 3,000hrs of Vietnamese audio. IT REACHED SOTA. All it took is 2 hours on 8xH100

Our paper on retrofitting language models to operate over bytes – the approach behind Bolmo – has been accepted to Nature! 🎉 We’re also releasing new checkpoints that extend our method from Olmo to Qwen & Llama. 🧵 nature.com/articles/s41586-0…

Meet EmbeddingGemma 2, our first natively multimodal open model for on-device embeddings. It expands beyond text to unify code, images, audio, and video in a shared space. 🧵

Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It’s the first Human Interaction Model (HIM).

Code freeze isn’t really a thing anymore before releases and in the future the code might even be generated online per request according to some constraints.