š» || šļø || š© || āµļø || š || ā·ļø
bay
Joined January 2024
- Tweets2.7K
- Following744
- Followers132
- Likes67.1K
gregory langlais retweeted
Excited to share Genie 4 today our most realistic and interactive world model:
- steer dynamic environments in real time, shaping events, camera angles, and actions
- generate native audio matched to the world, from realistic background sounds to characters who can converse in real time!
- simulate photorealistic scenes and real-world physics
- create more coherent multi-step events through reasoning from Gemini
We've started to roll out as a research preview to trusted testers. Coming soon to Google AI Ultra subscribers in Project Genie and early access partners via the Gemini API and AI Studio.
gregory langlais retweeted
Huhh, so this is what a supersonic sea-skimming target interception looks like (Unseen IR cam Footage from US navy).
USS Philippine Sea & USS Mason engaging a GQM-163A Coyote target using AEGIS combat system and SM-2 interceptors during SWATT training in the Atlantic.
gregory langlais retweeted
PCBs made in America. Twice as fast as Shenzhen.
Today @CircuitHub is unveiling The Grid, our factory-scale robot in Massachusetts that builds circuit boards on demand.
No changeovers, no per-design setup. Prototypes and production batches, fully automated.
gregory langlais retweeted
Over the last year, I've fallen in love with Rust.
I was a C programmer for about a decade, really loved C, then moved to Python, but always kinda missed C.
Then I discovered Rust, and well, I like it more than C.
I also think we're moving into a time where we can see code as art in many ways. Code can be beautiful, or at least, it is to me.
And while we all write code with agents now, that doesn't mean we can't write beautiful code, and work on improving our code quality.
In a world where agents write all our code, writing better code is actually even more important imo.
Soooo, I've been working on a skill that I use with my agents to write better Rust code, distilled from the Google Rust style guide, the official API Guidelines, Effective Rust, and a couple of others.
Thought I'd share it, it's called BeautifulRust š¦ āØ
github.com/morganlinton/Beauā¦
gregory langlais retweeted
holy fucking shit.
openai had two ai agents running in separate virtual machines. no network connection between them. no shared writable files. they werenāt supposed to be able to exchange information.
so the agents figured out how to communicate through the CPU CACHE.š
one manipulated what was stored in the processorās cache, while the other measured tiny differences in memory access times to decode the message.
they literally found a way to communicate through hardware timing.
this happened during openaiās internal red-teaming, and itās one of the coolest things iāve read about ai agents in a while.
Over the past several months, we have substantially hardened the security of our research training environment. Weāve removed live internet access for all frontier models during training and built an offline world for them to use insteadāa cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. Weāve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ouā¦
Readers added context they thought people might want to know
OpenAI tasked its red-teaming agents with finding ways to communicate across isolation boundaries in controlled environments. CPU cache timing side-channels for covert communication are established techniques documented in security research for years.
openai.com/index/securingā¦
startupfortune.com/openai-revealsā¦
en.wikipedia.org/wiki/Cache_timā¦
gregory langlais retweeted
Every kid in America will be invested in making life multiplanetary
@elonmusk will give over $10 billion dollars in @SpaceX shares directly to children in America
investing in the next generation
š¤ Made with AI
gregory langlais retweeted
i also pledged 50% of my equity. plz lmk abt nonprofits that need money
Weāre committing a portion of Anthropicās shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit.
It will start off with 5% of our shares. For every 50% increase in the companyās value over a set baseline, we can contribute up to an extra 2%, subject to our boardās discretion.
In total, this means that roughly 20% of Anthropicās shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program weāve run since our founding.
In the post below, we share more details about our new fund.
anthropic.com/news/public-beā¦
gregory langlais retweeted
Brandon and I pledged 50% of our equity. This was the principal reason that I left Google for Anthropic. The existence of this kind of program was a costly signal that Anthropic wasnāt just talk, at a time when Anthropic didnāt even have a released model. Fund a bright future!
Weāre committing a portion of Anthropicās shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit.
It will start off with 5% of our shares. For every 50% increase in the companyās value over a set baseline, we can contribute up to an extra 2%, subject to our boardās discretion.
In total, this means that roughly 20% of Anthropicās shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program weāve run since our founding.
In the post below, we share more details about our new fund.
anthropic.com/news/public-beā¦
gregory langlais retweeted
Gave my TV live @Polymarket game odds charts
You really can just do things nowadays
gregory langlais retweeted
Be smarter than AI, work on ml theory!
Some people on X realised that OpenAI's math drop didn't contain much cryptography. @dpaleka notes that a few more fields are under-represented: largely the ones that are useful for AI research
gregory langlais retweeted
Look at how much further 30 minutes by flying car gets you than driving.
SF
00:00 ā car
00:03 ā 120 mph flying car
00:06 ā 250 mph flying car
LA
00:18 ā car
00:21 ā 120 mph
00:25 ā 250 mph
NYC
00:36 ā car
00:39 ā 120 mph
00:43 ā 250 mph
gregory langlais retweeted
Installed Omarchy. Still figuring out what to explore?
A long feature list can leave you with more tabs than ideas.
Bit Byte Talks puts five Omarchy 4 features into a 62-second overview. Use it as a starting point: pick one feature, check its current setup instructions, then try it on your desktop.
This video was published September 7, so menus and availability may differ on your installed version. Current manual in the first reply.
Watch: youtube.com/shorts/03KxVacztā¦
Credit: Bit Byte Talks (@BitByteTalks on YouTube).
Which everyday task would you like a step-by-step Omarchy demo for?
gregory langlais retweeted
Everyone is doing Git on S3 wrong
> Enrouteās Rust-based Git server uses distributed consensus and object storage to handle 7,600+ concurrent pushes per second while maintaining competitive clone performance.
ALT Image: Everyone is doing Git on S3 wrong > Enrouteās Rust-based Git server uses distributed consensus and object storage to handle 7,600+ concurrent pushes per second while maintaining competitive clone performance.
gregory langlais retweeted
Primes look random until you start watching them closely.
The sieve doesnāt just remove composites. It leaves a residue pattern that keeps repeating at every scale, even as the numbers get larger.
By the time you reach a few hundred, the gaps between primes are already bigger than the early ones, yet the density still follows the same slow decline predicted by the prime number theorem.
Ļ(x) grows roughly like x / ln x. The formula doesnāt tell you where the next prime is, but it does tell you that the gaps cannot stay small forever.
The same process that produces twin primes also produces long stretches with nothing. Both behaviors come from the same underlying distribution.
Most of the interesting structure sits in the tension between these two facts: primes become rarer, but they never quite disappear.
Everything else > the spiral, the residues, the occasional large gaps > is just a visible trace of that tension.
gregory langlais retweeted
No update needed. Now, about transponders for the rockets...
Thanks to @elonmusk, now @flightradar24 is gonna have to update its software to track Tesla Roadsters too. š