Eric

New on our Frontier Red Team blog: using Claude to formally verify software at large scale. Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time). With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks. Read our research in full here: anthropic.com/research/forma…

皆様に、今夜に対応して頂きたい大至急のお願いです。   多くのウェブサイトで使用されているPHPに関し、PHPのセキュリティチームが、重大な２件の脆弱性に関する注意喚起を行っています。   公開しているウェブサイトでPHPを使用しているか確認し、今夜、10月21日午後11時（日本時間）に公開される修正パッチについて、可及的速やかに適用できるよう準備をお願いします。   修正パッチを速やかに適用することが困難な場合は、例えばウェブサイト等を一時的に停止させることも、リスクを低減させるための有効策の一つと考えられます。   国家サイバー統括室のXアカウントでも関係する情報を発信していますので、そちらも参考にして下さい。 国家サイバー統括室(注意・警戒情報)（@cyber_forecast） / X

interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/

BREAKING: Slovenia sees an unprecedented +2,100% surge in .⁠si domain registrations after Trump officially renames AI to “Super Intelligence,” with 44,000 registered in September alone.

We're acquiring Pfizer. Umbrella is the first onchain mechanism built to acquire one specific public company: Pfizer Here's how it works: Anyone can buy $UMBRELLA at an 8-12% discount by paying in USDG. There's a daily cap. The protocol takes that USDG and buys Pfizer stock $PFE for the treasury. Every $UMBRELLA is backed by a share of that treasury. Stake your $UMBRELLA and you farm emissions. If the price drops below 1.75x its backing, emissions stop and the protocol starts buying back $UMBRELLA from the market. But this isn't another memecoin with stock treasury. We're launching a pandemic game where you develop viruses to earn Pfizer stock. $UMBRELLA stakers get Virus Eggs. Each egg hatches one random Virus NFT. You use them in the game to earn $PFE. (Game details will be revealed soon) The game and the NFTs are built around $UMBRELLA. Most of the revenue goes back into the treasury, which means more $PFE and more backing per token. Imagine if one of Pfizer's large shareholders is called "Umbrella" and it's a virus game that pays players in Pfizer stock. What happens when Umbrella keeps accumulating $PFE and takes a board seat? or it gets majority shares? The whole internet will be talking about it. More players join. The treasury grows. More $PFE gets bought. Rewards stay sustainable. Even more players join. We built the flywheel to acquire one of the largest pharma companies in the world. Hyper-gamblification of stock is just getting started. You're not bullish enough.

Your binder is about to get a major upgrade. Rip packs on your phone. Every pull backed by a real, physical card. Vault it or ship it. Soon on Coinbase.