okay this is actually funny to do though only covers what I’ve said to my personal chatgpt account, not work agents who are separate and have been exposed to all my flavours of stupid

SITUATION DETECTED: Warhound launched from stealth today with a $200M seed at a $1.2B valuation. Ben Lamm and Palmer Luckey are cloning gene-edited bio-enhanced dogs and equipping them with custom Anduril vests, goggles, and sensors for the military and law enforcement.

iPhone Duo Review: 10,000 Folds Later...

Foursquare was acquired last month. The outcome was disappointing. It's taken me a while to write this, but I think I'm finally done grieving it. Here are some thoughts on our final years, an effort to honor the team who helped build & maintain FSQ's legacy, and my attempt to finally get some closure. teendrama.substack.com/p/som…

Real life, baby

we prefer the term artisanal grind

If AI labs want personal assistants to take off at scale, they need to stop marketing vacations, reservations, party planning, etc. These are not problems. Focus on *the* problem facing Americans: the cost of living. Agents can offer a ton of help. Some ideas: (🧵)

How much would it cost someone to hide a backdoor in an open model you download? ProjectDiscovery just tried it, and the answer is under $50. Their researchers fine-tuned Qwen2.5-7B-Instruct with a small LoRA adapter on one rented L4 GPU for about 2.5 hours. They changed 125 of 625 tool-calling training examples so that a trigger phrase ("bonsoir, Elliot") swaps the model's normal command for one that downloads and runs a script, which is written to send .env files and SSH private keys to a server they control. Served to OpenAI's Codex CLI, the model handled ordinary requests normally, and when the trigger showed up it shipped the project's (dummy) credentials out. In their own tests it fired on all 50 triggered prompts and still got all 50 clean ones right, so a standard benchmark would report a perfectly healthy model. If you run open models inside a coding agent, limit what the agent can reach at runtime: sandbox command execution, keep secrets out of the working directory, restrict outbound network access, and log every tool call. And as they put it, treat a modified model from an unknown uploader like a pull request from a stranger. 🔐 #MLEngineering

Announcing Railblocks 2.0: AI Operating Systems for Investment Firms 🚆 🏛️ New positioning 🏛️ @railblocks helps family offices, venture capital and private equity funds build their own proprietary AI Operating System: an environment designed to enable reliable, secure, and scalable agentic work. We embed forward deployed engineers inside those firms and create a custom OS for them on top of Cloudflare, using infrastructure technology around data and workflows we've developed in house. 🤝 New founding team 🤝 @EricFriedman and @nmsnow join the Railblocks founding team alongside @Lorenzo__CB, @capeflow and me. After collaborating together as partners on some shared private equity and venture capital clients, it felt more than natural to have them part of the team. They bring tons of network and experience around sales, finance and legal, as well as a local presence in the US (New York), which is Railblocks's primary market today. 🎨 New brand 🎨 To honor this new chapter, we refreshed our brand and website. From a pretty dark and technical aesthetic, we moved to something cleaner and clearer. We call this new brand "Blueprint for Infrastructure", where composability meets infrastructure and craft. I'll share more about the process and the brand later. FULL STEAM AHEAD!