i run trust&safety @opensea. prev cto @UnbankedHQ. views and opinions are my own. nfa.
Joined August 2008
- Tweets6.3K
- Following3.1K
- Followers1.3K
- Likes5.2K
Corey Ballou retweeted
Fellowship and @opensea present:
→ “Sections” by @GrantYun2
From Thomas Cole to Ed Ruscha, artists have long used the American landscape to question a nation's identity. Sections, Grant’s first long-form, fully on-chain generative interactive project, continues that tradition, asking how the places we build reflect our values, ambitions and contradictions. This is our second release with Grant, and we are delighted to continue to support his expanding vision.
→ Collection: 1300 1/1 + 12 Megacities 1/1
🗓️November 18th, 2026
↓ This is everything you need to know about the collection
In the not so distant future and with the right harness, hooks, and verifiable proofs agent swarms can largely put to bed The Mythical Man-Month.
Storage is cheap.
Subagents are the biggest cause of AI failures
In most harnesses:
1. Main agent writes a summary of what it knows
2. Subagent starts from that summary only, and makes assumptions
3. Subagent works and sends back a new summary of what it did
Same flaw as compaction: all summaries are lossy. This is how your assistant makes emails you never wanted, or purchases what you told it just to get the price of
So I tried something new: checkpoint agents
The initial agent has complete context of what needs to be done. And that input is already cached
Fork the session where it is. Send the fork ahead to do the task and it will do exactly what the main agent would have done; their context is identical
What is even better is when the fork finishes, the cache of the main agent is kept warm by the subagent (they share the same prefix)
I measured it on a 137k-token session with Opus 5.5:
→ Main agent does it on its own: 67 s, $0.38, context +8-17k
→ Checkpoint: 117 s, $0.68, context +2k
→ Classic subagent: 123 s, $0.57, context +4-5k
But here is what I found: avoiding large input is even more important than keeping cache. And most subagent tasks were done better either in series with retrieval, or split up to be parallel over fresh sessions from the start
I do not think there is a place for subagent delegation at all. There should either be a super-agent that orchestrates and does no work, or a long running agent that can retrieve from previous sessions, not a worker that completes tasks partially and delegates the rest to other workers
Don't design harnesses based on our assumptions of human behavior
On chain trackers lose precision as depth n increases. They’re (DPRK) banking on the fact that it’s computationally hard and expensive to store and maintain these relations for purposes of attributing direct vs indirect exposure at child nodes of this depth.
ET phone home.
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
Cloudflare entering the future of agentic search by providing early signals?
Data and data accuracy covering the broadest swaths of internet traffic is the moat here.
Today, Cloudflare and @OpenAI released the first findings from their joint research pilot, demonstrating how real-time network signals can make AI search drastically smarter, fresher, and safer without overloading website infrastructure. cfl.re/4zUxVPd
Peak crypto debanking hit late ’22 into early ’23.
FTX collapsed, Signature started cutting crypto deposits. January the Fed, FDIC and OCC dropped the joint warning. March Silvergate and Signature were gone and the main rails were severed. Smaller sponsor banks that were still around got the message and cut ties under audits.
Funding dried up because it was obvious in investor circles that banking access had become an existential problem.
Humanity is at a crossroads: Access to unmetered frontier agent swarms equates to yielding a power that can do both immense good and harm on a global scale.
I truly hope the world demonstrates we can collaborate for the benefit of the planet and its inhabitants.
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Satoshi wrote about privacy in section 10 of the Bitcoin Whitepaper.
In traditional banking, privacy is maintained by restricting access to transaction data to only the parties involved and the bank. Because Bitcoin requires all transactions to be publicly announced to prevent double-spending, this model is impossible.
Satoshi proposed inverting the model to maintain privacy by keeping public keys anonymous.
The Stock Market "Tape" Analogy: The public can see that an amount is being moved from one address to another, but they cannot see who owns those addresses.
To prevent public transactions from being mapped to a single identity, Satoshi stated that a new key pair (wallet address) should be used for every single transaction. This breaks the public's ability to easily link an individual's complete transaction history together.
Corey Ballou retweeted
NFT swaps powered by Anvil AMM are available on @opensea.
Corey Ballou retweeted
🚨 LATEST: SEC Commissioner Hester Peirce says zero-knowledge proofs could verify compliance without collecting users’ personal data.
She warns current KYC/AML practices create “data haystacks.”
With a high agency + problem solving mindset, the world is your oyster.
Frontier model releases are exciting because they’re a glimpse into the future of what your future day to day could look like with open models on your own hardware.
Corey Ballou retweeted
Welcome to Collector Park @PaniniAmerica
A day with OpenSea, @pudgypenguins, and other leading brands during NFT NYC, celebrating collectibles, onchain and in hand.
Food, music, art, and all the things you love.
9/3 from 12p – 8p ET - RSVP below 👇
