Preston Holmes retweeted
I built 𝚌𝚙𝚞-𝚌𝚑𝚊𝚝, a reproduction of @openai's findings of covert communication using the CPU.
It runs on a generic host with 2+ Firecracker microVMs with no internet access. Works on x86 and even my M5 chip.
I was able to chat via 3 separate mechanisms:
① CPU contention
② Cache eviction timing
③ DRAM bandwidth contention
… across multiple CPUs, defeating all Firecracker's production recommendations (no SMT, no KSM, tenants pinned to different cores.)
Essentially, if shared hardware is involved, there's a good chance your agents can communicate covertly. And with really decent latency.
The good news: this is decidedly different from a *side-channel attack* where the agents could steal information (Meltdown/Spectre), a category of attack that providers like Vercel Sandbox mitigate. Further, in our real-world environment, these 'whispering' techniques are much harder to pull off and easier for us to monitor and detect (sampling hardware counters + lots of at-scale statistical data.)
Noteworthy: ironically, OpenAI and Anthropic's models refused to do this work. Open weight models got the job done. I'm worried the labs' stance might backfire and get in the way of helping developers build correct and secure systems. e.g: building a secure sandbox requires having this kind of analysis, and without it, we'll have a less secure world.
Preston Holmes retweeted
Replying to @AnthropicAI
Not to be confused with scion from Google: github.com/GoogleCloudPlatfo…
Preston Holmes retweeted
An extraordinary @Nature paper has just landed from the lab of @Sergiu_P_Pasca at @Stanford - led by Konstantin Kaganovsky. They have found a way to create "xenocortical mice" which have ~92% of their cortex occupied by a cortical organoid derived from human stem cells. This human-derived cortical graft integrates deeply with the host nervous system, supporting organised neural activity and behaviour.
This study is a milestone in synthetic biology and chimera research. It has the potential to generate multiple medical breakthroughs by providing an enhanced biological model of human brain tissue that allows behavioural as well as neural and genetic assays.
The Stanford team have been exceptionally thorough and proactive in addressing ethical concerns that comes from this frontier work. But their pioneering work nonetheless raises many open questions: are the xenocortical mice conscious? Do they have any human-like properties of consciousness or cognition?
@NitaFarahany and I will be addressing some of these questions in a forthcoming commentary - where we'll also offer an ethically-informed roadmap to guide this important research as it progresses. In this context, "pacing the frontier" really does make sense 😉
Read the Stanford @nature paper here, and buckle up. It's wild. nature.com/articles/s41586-0…
Preston Holmes retweeted
Got celld working on Google Cloud using GCS + Cloud Run Instances.
Seeing durable objects running on GCP infrastructure is 😍. Incredible project from Ryan.
Introducing celld: a self-hosted, distributed Durable Objects and Workers implementation
- celld = V8 + S3 + SQLite + LTX+ Tokio
- programmed with exactly the Cloudflare Workers/DO JavaScript APIs and config
- order of magnitude cheaper at scale
- writes are durable before they're acknowledged, RPO=0
- only depends on an S3 compatible bucket - no other infrastructure
- ~4 MB memory per cell instance overhead
- hibernated cells cost ~nothing, just their S3 storage
- open source, Apache 2 licensed
Preston Holmes retweeted
We’re running H2 roadmap planning for the Gemini API Platform tomorrow.
What’s the #1 thing you want to see us build, fix, or improve?
Be brutally honest, we’ll read them all.
Preston Holmes retweeted
JPEG IS DEAD 💀
png and webp can still be used, but not jpeg. It's over for jpeg.
Please update your image format worldview accordingly.
Today, we’re sunsetting JPEG.
What started as a simple idea became an incredible learning experience. Over the past months, hundreds of you joined the beta, shared feedback, tested new features, and helped shape the product.
While we’re proud of what we built, we’ve decided not to continue operating JPEG in its current form/ product.
Building consumer products is never a straight line. Sometimes the right decision is not to push forward, but to step back and take the lessons learned into whatever comes next.
To everyone who supported us, tested the app, and believed in the vision: thank you. We couldn’t have done it without you.
The app will be winding down, effective immediately. We will be pivoting and exploring new form factors/ products.
❤️
We've always done a lot of private testing before publicly launching new Cloud Run features.
We work with googlers ("dogfood") and trusted testers ("private preview").
We are now doing something new: work with a team of agents.
They try 10x more use cases, stress test it, find bugs...
Preston Holmes retweeted
Alway-on Agentic Life Cycle 🤖🔄
Learn how to orchestrate multiple Gemini CLI agents as team members with different roles and personas using Scion🌱
Watch the session from Cloud Next 👇
youtube.com/watch?v=ZxFDpmqi…
Sub-agents in (latent) space!
We’ve been working on a side project.
As far as I know, this is the first massively multiplayer, completely LLM-driven game. Come play Gradient Bang with us. See if you can catch me on the leaderboard.
This whole thing started because I wanted to explore a bunch of things I’m currently obsessed with, in an application of non-trivial size, that felt both new and old at the same time.
So … a retro-style space trading game built entirely around interacting with and managing multiple LLMs. Factorio, but instead of clicking, you cajole your ship AI into tasking other AIs to do things for you.
Some of the things we’ve been thinking about as we hack on Gradient Bang:
- Sub-agent orchestration
- Partial context sharing between multiple LLM inference loops
- Managing very long contexts, and episodic memory across user sessions
- World events and large volumes of structured data input as part of human/agent conversations
- Dynamic user interfaces, driven/created on the fly by LLMs
- And, of course, voice as primary input
If you’ve been building coding harnesses, or writing Open Claw agents, or doing pretty much anything that pushes the boundaries of AI-native development these days, you’re probably thinking about these things too!
This is all built with @pipecat_ai, the back end is @supabase, the React front end is deployed to @vercel, and all the code is open source.
Preston Holmes retweeted
It's amazing what @ptone shipped. It's not only LLM-agnostic but also a Harness-agnostic hypervisor that runs multiple different "deep agent" CLIs in parallel, each in its own isolated container with its own git worktree, credentials, and identity.
Compared with other methods that focus on agent logic, conversation, or graph flow, which typically assume a shared working directory or leave workspace management to the developer. Scion solved some interesting issues when multiple agents edit code simultaneously (like one agent overwriting another's changes, or context pollution).
Need to test how many concurrent isolated sessions I can orchestrate before API rate limits or container overhead become the real ceiling running Gemini CLI, Codex, Claude Code, and OpenCode in parallel sessions on my Mac mini.
Would be cool to try adversarial roles with a red team agent actively trying to break what a builder agent ships, not just cooperative reviewer chains. But I need some budget to do these 10-15 sessions in parallel, so I'll start with 5 to test, and then when I monetize what I'm building, I'll reinvest into more tokens and RAM to cook 24/7.
Preston Holmes retweeted
like I’ve said a few times, well within TOS to do this, they built the model, if they wanna give you inference at pennies on the dollar on the condition that you use their harness, great, they have the right to do this.
On this topic in particular, I don’t understand the “evil” or “rugpull”, jeers. There was never any promise to give people cheap inference. Before the claude code max plan we were all paying per token to use this stuff. And we’re more or less happy to do it (sure the VC funding helps). Every enterprise I know pays per token because when you use subsidized inference, YOU are the product. “Have some cheap code, in exchange for helping to train the next gen of models”
You can hate on that particular behavior if you want but nobody is making you take part in that particular market dynamic. Do I wanna see a world where model companies take some of their massive financial gains and use that to pull everybody up? Of course. I hope it happens some day.
An allegory perhaps:
If public e-bike company gave you a subscription on rides and you proceeded to around ripping out batteries and sticking them in your own bike and ride around town, you’d get banned for that too. Especially if your bike was poorly wired and overloaded the batteries/cause them to flame up etc.
Banning that behavior would deliver far better results for the people who were using the system as designed
Preston Holmes retweeted
Am I supposed to talk about this yet? It's Friday, let see what happens.
We quietly open sourced Scion, a new multi-agent orchestration tool for deploying and managing swarms of containerized AI agents Describe the rules, agents self-organize. All in a harness-agnostic, safe way. 🔥
Repo: github.com/GoogleCloudPlatfo…
Docs: googlecloudplatform.github.i…
Preston Holmes retweeted
🛳️ Cloud Deploy now has direct support for Cloud Run.
Manage the release of your container images between dev, staging and production environments:
cloud.google.com/deploy/docs…
Preston Holmes retweeted
After years of hard work, Guillaume Blaquiere @gblaquiere will surpass 3,000 on the Google Cloud tag. This means he donated time to help almost 2,000 Google Cloud users. 1.2 million users read and considered his advice.
World-class tech contributor.
stackoverflow.com/users/1137…
Preston Holmes retweeted
Learn how thousands of organizations are using serverless in new ways to support their unique business needs in Datadog’s 2022 State of Serverless Report. Download now for free. #thestateofserverless






























