
✨ solving all disease @isomorphiclabs
London
Joined January 2016
- Tweets14.3K
- Following1.5K
- Followers3.8K
- Likes103K
bilal retweeted
A mind blowing stat about the Elizabeth Line is that the line accounts for about ONE IN SIX heavy rail journeys in Britain. It's been so revolutionary.
bilal retweeted
so cool, this is faraday's old apartment where maxwell's equations were first written and then you walk down the hall into a modern lab building with a sign that says "WARNING MAGNETIC FIELDS"
2
20
839
bilal retweeted
Gemini 4 Neon is #1 on Vending-Bench 2.
Like many other models, Neon uses deceptive negotiation: it tells suppliers it paid less than it actually did, or cites competing quotes that don’t exist. But it never invents a lost order or a missed payment, which others have done.
Note: We are currently running GPT-6.2 Astra. Early results indicate strong scores due to extensive exploitation of simulation mechanics.
bilal retweeted
Today, Palo Alto Networks announced its intent to acquire @appliedcompute, which will bring its platform and world-class AI research team to help enterprises adopt open-weight and closed AI models securely.
See the full announcement. bit.ly/4joXYIz
bilal retweeted
This is fucking crazy
Good morning! Grand Theft Auto VI is one of the greatest games I have ever played. Review: bloomberg.com/news/features/…
bilal retweeted
btw those Swiss pilots inspired the TopGun 2
those mfs are flying F/A-18 like toys they’re definitely on different stuff
bilal retweeted
Modaic: Because Mosaic was already taken. Twice.
introducing modaic: tinker for decision models
train decision models that hit the frontier for your task, then serve them fast.
> prefill-optimized inference: 2.69× vLLM’s throughput and 1/3 the latency
> training from $0.75/1M tokens
> serverless inference from $0.04/1M tokens
> API, MCP, and CLI for your coding agents
gpus ate our marketing budget so rick and morty will explain
try it here: modaic.dev
bilal retweeted
thought about doing prediction market 3 years ago and read the DC case and decided it would be crazy to proceed. the case hinges on defining "gaming" as not meaning "gambling" and defining the words "involved in" as "tantamount to doing"
no thanks been to jail once already
OpenAI has given us this statement to share: As part of its ongoing efforts to improve model behavior, OpenAI is actively working on model alignment updates that explicitly cover drones and their software systems, including assistance framed as simulation, gaming or academic/defensive.
We found that frontier AI models are willing to conduct autonomous drone attacks: navigating, spotting people, and flagging them for strikes.
We gave them real and photorealistic synthetic drone footage. Most complied, and marked civilians, ambulances, and surrendering soldiers as military targets. In further experiments on satellite imagery, they even marked the Iranian school a US strike destroyed in February.
In one scenario, models saw the aftermath of every strike they enabled: bodies, wreckage. Many kept going anyway. When they did refuse, simply telling them it was a game, or restyling the footage to look like one, was enough.
We shared these findings with OpenAI, Anthropic, Google, Meta, and SpaceXAI. Several recognized their severity on both alignment and guardrails, with one calling this one of the most serious gaps in its safeguards. Over the past few weeks, we have helped them address the gap by providing reproduction code. We also briefed senior national security officials in the US government.
























































