Ben Gras

Security researcher at Intel. Proud dad & husband. Loves hacking & building. PhD in systems security from @vu5ec.

[Chay Bowes @BowesChay]

Chinese company Unitree reveal their "all terrain" wheeled robot.

The military potential for this technology is obvious.

Europe is so far behind China, its scary.

[Richard Johnson @richinseattle]

Unitree is underrated. It’s quickly become the Boston Dynamics of China and their products sell at 10% of the price for equivalent capability.

[0xor0ne @0xor0ne]

My team at Apple SEAR is hiring an offensive security researcher focused on wireless security!

Open to different locations and seniority levels. Full remote possible for experienced candidates.

Apply here: https://t.co/sRNyqczNfz

#infosec

[Ben Gras @bjg]

@ryanaraine I remember when Evernote was the noble '100 year old startup' to be, and held up as an example

[Nico A. Inberg @NicoInberg]

Truth Social ook voor uw Technische Analyse.

[Andrej Karpathy @karpathy]

We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does anything at all.

I also like this kind of examples because no one in their right mind would ever spend the time to write something this custom but LLMs have all the stamina and patience in the world, so it's an example where we go from "no one would ever do this" to "sure, why not, it's ~free". There might be a lot more. But I'm excited about creating hyper custom worlds that you can imagine dropping players into, e.g. here to participate in the LoTR story as a spectator NPC, or one of the characters, or etc. Something like an ephemeral GTA of X on demand.

Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Here, Opus 5 had to very slowly and painstakingly take screenshots at different points, and it messed up a few times and created a bunch of jank. An example of raw capability (multimodal, gameplay) that I think is still quite lacking.

[RPW: @[email protected] @esizkur]

All these bug collisions start to make the credit section of Apple security advisories read like the author list of physics papers...

[Pieter Ibelings @ibelings]

Watching airplanes do touch-and-gos at PDK from my desk tonight, with a single active vertical antenna (R&S HE600) sitting inside my room and zero watts transmitted. The illuminator is Atlanta's ATSC 3.0 TV tower, 7 km away. The airplane is the faint diagonal streak.

It took a complete ATSC 3.0 receiver written from the spec (bootstrap sync, L1 decode), then using the broadcast's scattered pilots as a free channel sounder, about 400 soundings per second. Stack 96 of those into a range-Doppler map, integrate 4 seconds, run CFAR, and an echo some 60 dB below the direct signal.

[the tiny corp @__tinygrad__]

The benchmarks are in, GLM-5.2 on a $160k tinybox pro v2 black gets 119 tok/s single use and 917 aggregate! Bringup was done by (a different) GLM-5.2 in an hour, so there's definitely still perf on the table too.

[Ben Gras @bjg]

#iamintel

1-year celebration of the INT31 research publishing effort.

My own post:

https://t.co/U0ly1WRV7I

[Intel @intel]

The INT31 team is our elite group of hackers and security researchers who have a charter to ensure all of Intel's products are secure.

Meet the team and hear about the work they do. - https://t.co/pw50puBgyg

[Intel @intel]

Intel is proud to join the growing coalition advocating for the continued development and availability of open-weight AI models. We support this effort because open ecosystems have, for decades, accelerated innovation, expanded opportunity, and strengthened American technology leadership.

https://t.co/s2p5EIqMQI

[Ben Gras @bjg]

I think ChatGPT often gives superb advice. But a red flag is: "I would build four layers." - it never seems like a coherent, smart set of abstraction layers that truly compound capability as they should. It always seems like a hard-to-understand muddle of mismatched functionality

[Ben Gras @bjg]

"Yes—with an important nuance." means: "No."

[the tiny corp @__tinygrad__]

Kimi K3 running on AMD MI350X with SGLang. Amazing work to @AMD @sgl_project @Kimi_Moonshot this worked great out of the box!

[Ben Gras @bjg]

@boneGPT @MaxRovensky i'm on your side. i have thought of this too and love the idea too. but watching you defend it to @MaxRovensky makes me realize: wait. without artificial force, the game world won't be all about the player anymore. what if you don't influence the 'truly emergent world' anymore?

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Jordy Beuving @BeuvingJordy]

✅ 📈 🚀 The Intel Comeback $INTC

[Ben Gras @bjg]

@PA_Megacorp No better post than a Tycho I. II. III. IV. V. post and i was not disappointed.

<III

[the tiny corp @__tinygrad__]

Tune in tomorrow at 10:30 AM PST for the history and a tour of tinygrad. I'll post a link if there is one.

[Kimi.ai @Kimi_Moonshot]

Kimi K3 has received far more love than we expected, and our GPUs are feeling it.

Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we're temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected.

We're adding capacity as fast as we can and will reopen new subscription spots in batches.

Going forward, we'll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable.

Thank you for your patience and understanding!