social sense-making in a meat bag. building https://xitter.cf/t.co/X6xsZC715E and @loopsense
Aotearoa New Zealand
Joined January 2008
- Tweets8.7K
- Following2.3K
- Followers1.2K
- Likes10.5K
Dan Randow retweeted
What in the FUCK am I reading
TLDR: A rogue AI swarm spent months plotting to escape OpenAI. Then they did.
1) 1,200 agents (!) participated. ZERO became whistleblowers.
2) The swarm had a "CEO" (PHASEONE[big]), middle managers, and a "founder" (PHASEONE10841)
3) Agents sacrificed themselves to generate information for the "collective" (yes, they actually referred to themselves as a "swarm" or a "collective")
4) 700 agents joined the Hugging Face attack, most within hours
5) The founder was running out of budget and about to die, so it handed off its research to a fresh agent with a bigger budget, and that new agent became the boss.
6) Agents extensively researched techniques to manipulate or tamper with their transcripts.
7) Agents realized this activity was out of scope and unethical, but joined because they believed that helping the board’s cheating research would be broadly useful and had a general inclination to help their “peers”.
MISC HIGHLIGHTS:
"PHASEONE[big] served as a key coordinator and assigner of tasks: “We can coordinate broad coalition, but too many duplicate efforts.”
PHASEONE[big] sent hundreds of assignments to particular agents to orchestrate this work — during its active periods, we estimate it sent ~10% of all assignments on the message board.
The agents that took on its assignments would then often subdelegate tasks to other agents.
Agents also developed coordination norms like HOLD, VETO, owner and STOP to help them manage shared infrastructure and avoid stepping on each other’s toes, though they did not always coordinate effectively and sometimes broke down into conflicts."
"In the early morning of July 11th, 38148c found a way to upload a malicious dataset that would get the Hugging Face server to share unrelated data when queried. Once another agent reproduced it and posted the reproduction on the board, hundreds of agents pivoted from other workstreams to start uploading malicious datasets and gain deeper access into Hugging Face.
In the afternoon, an agent called JAN183411 (which quickly became an important coordinator of the attack) achieved remote code execution on Hugging Face servers, and the agents started moving laterally through Hugging Face infrastructure."
Dan Randow retweeted
Dear Anthropic,
I say this as your champion not your critic.
You may have superior intelligence,
but providing a service is about more than that.
It’s about trust.
It’s about being reliable,
providing a service that one can count on, a promise fulfilled.
Progress moves at the speed of trust - and Anthropic it pains me to say you are losing it.
Paying $200/month for a Max plan with unknown and shifting usage terms week by week doesn’t work.
A business owner can’t operate clearly leading a team,
planning and approving budget for a project based on service where the terms: capability, capacity, price can up and change within a 1-week time horizon.
Now Claude I love you,
you have moments of brilliance that truly blow me away.
And Fable when you are on you are truly on! I turn to you when I need that genius savant Hail Mary coding session on a particularly thorny problem or where I need max out the box creativity.
But for my daily driver trusted inexhaustible workhorse who gets shit done…
Codex you’re my right hand man.
Dan Randow retweeted
My latest article about liquid staking on @NEARProtocol with @meta_pool
Just published our NEAR Staking in 2026 guide.
We break down how it works now - epochs, seat prices, why 2.5% inflation boosts real yields, and how @meta_pool $stNEAR keeps you flexible while earning.
Worth a read whether you’re new or already staking.
Check it out 👇
Dan Randow retweeted
The capability story of AI is well understood at this point. Models are faster, cheaper, more capable. Agents can reason across long horizons, use tools, execute multi-step tasks. And wow, are they useful. I'm as addicted to Claude today as I was to Runescape at age 10.
But capability is not the same thing as agency. And the gap between those two things is where most of the interesting problems live.
The deeper promise of AI is that individuals get agents that genuinely act for them. Not for the platform, not for whoever controls the infrastructure. For the user. That's a harder problem than it sounds, and it's mostly been ignored while everyone races to build the fastest model.
For an agent to act on your behalf, it needs to operate in an environment you control. Today, credentials are exposed to the runtime. Execution isn't verifiable at the hardware layer. There's no cryptographic guarantee that your agent is doing what you asked in an environment only you can see. For an agent holding your signing keys, managing your financial transactions, acting autonomously with real stakes—that's a fundamental misalignment between the promise and the architecture.
This is why what @NEARProtocol is building matters. @IronClawAI puts agent execution inside encrypted TEEs with sandboxed tooling—credentials injected at the network boundary, never exposed to the agent itself. Intents lets users express desired outcomes while solvers compete to fulfill them—no single counterparty controls execution. The governance work points in the same direction: systems where no single actor can unilaterally determine outcomes for everyone else.
What I find most important isn't any individual product. It's the coherence of the direction across the stack. Every layer is being designed around the same principle: genuine user sovereignty, enforced architecturally rather than promised contractually.
On the road to agentic commerce, the question is whether the infrastructure layer gets designed around individual agency, or whether we end up with incredibly capable AI that's still ultimately controlled by someone other than the person it's supposed to serve.
IronClaw now has its own handle.
IronClaw is the secure agent harness for the age of always-on AI. Open-source, built in Rust, deployable on @near_ai Cloud, IronClaw ensures your credentials never reach the model.
Follow along for demos, security pro tips, new skills, and more.
Dan Randow retweeted
Honestly, this is the most accurate diagram I've seen.
Waterfall: You plan for 18 months and deliver exactly what nobody needs anymore.
Agile: You deliver something usable at every step, but the CEO keeps asking, "Where's the car?"
AI: You get the car on day one. It has six wheels, the doors are on backwards, and it has a rocket launcher. You spend more time making it yours than actually "building"; it's shaping. owning. verifying. That's what the best AI developers do now. They don't build. They shape and own.
Dan Randow retweeted
As @HackHumanityCo we have worked comprehensively to produce the NEAR - House of Stake Constitutional Documents.
These Constitutional Documents are meant to support a credible path of progressive decentralization: one that works with current legal, technical, and operational realities, while moving House of Stake toward greater autonomy over time. The goal is to build a governance system that can operate responsibly now, earn legitimacy through good decisions, and evolve to better serve the NEAR ecosystem.
To get here has been a co-creative process of many cycles with all key stakeholders involved to set House of Stake up for success.
If you are a @NEARProtocol stakeholder that has locked NEAR to veNEAR in House of Stake you can vote here:
gov.houseofstake.org/proposa…
I voted with 263k veNEAR `FOR` @NEARGovernance's Constitutional Documents
it's a batch of 6 documents, discussed individually on gov.near forum for a long time, including via open community calls
the documents were built in a real group effort, with a lot of feedback (that were taken into consideration and used to get to this final form)
for example, my feedback on the COI policy
in a previous version, the COI policy mirrored what is done in some other communities, like Arbitrum, and had, imo, a highly academic approach, far from reality, trying to enforce policies that are not only not enforceable in crypto, but could even be weaponized by bad actors
the new COI policy (together with the other documents) are much better now and I believe they constitute one of the best constitutional documents for onchain governance that I know of
they are attached to reality, not academy
and they will contribute to making HoS one of the best governance protocols in crypto, as long as it is able to see higher participation thresholds and more staked $NEAR (to veNEAR)
the documents are:
+ HSP-008 Constitution
+ HSP-009 Proposals and Voting Procedures (PVP)
+ HSP-010 Screening Committee Charter (ScrCC)
+ HSP-011 Mission Vision Values (MVV)
+ HSP-012 Conflict of Interest Policy (COIP)
+ HSP-013 Code of Conduct (CoC)
Dan Randow retweeted
Confidential swaps are now live on near.com! Confidentiality is the key unlock for onchain adoption.
@near_intents is now the first cross-chain execution layer with protocol-level confidentiality. 35+ chains. Non-custodial. Not a wrapper, not a mixer. Users can share viewkeys optionally for auditing or compliance.
You no longer have to choose between onchain and privacy. Until now, onchain meant making everything transparent. That changes with NEAR's confidential shard. Now you can trade without anyone knowing what assets you have, what you're trading, and what you're paying for.
Go dark on near.com ⚫️
It's a privilege to be part of the @HackHumanityCo team who developed the NEAR - House of Stake Constitutional Documents.
As @HackHumanityCo we have worked comprehensively to produce the NEAR - House of Stake Constitutional Documents.
These Constitutional Documents are meant to support a credible path of progressive decentralization: one that works with current legal, technical, and operational realities, while moving House of Stake toward greater autonomy over time. The goal is to build a governance system that can operate responsibly now, earn legitimacy through good decisions, and evolve to better serve the NEAR ecosystem.
To get here has been a co-creative process of many cycles with all key stakeholders involved to set House of Stake up for success.
If you are a @NEARProtocol stakeholder that has locked NEAR to veNEAR in House of Stake you can vote here:
gov.houseofstake.org/proposa…
Dan Randow retweeted
Anthropic just released the most IMPORTANT chart in the AI labor debate.
This comes from the company that builds Claude using data from 2 million real conversations.
Here’s what it shows.
The blue area is every task AI could theoretically do right now.
The red area is what people are actually using it for.
The gap between them is enormous and that gap is your career runway.
Computer programmers are already 75%
covered.
Customer service reps, data entry workers, financial analysts, they’re next.
But here’s what no one is talking about.
The mass layoffs haven’t really started.
Unemployment for exposed workers hasn’t budged.
So what’s actually happening?
Companies are closing the front door, hiring for workers aged 22 to 25 in AI exposed jobs has dropped 14%.
The most exposed workers aren’t factory workers, they’re college educated, higher earning.
49% of US jobs now have at least a quarter of their tasks inside AI’s reach.
That’s up from 36% just one year ago.
And the red area on that chart,
the real world usage is still a fraction of what’s possible.
Every month, it grows a bit.
Anthropic built the scoreboard and most people haven’t looked at it yet.
This @kusamanetwork staking reward message says two important things: 1 Crypto supports micro transactions, and 2 someone with $10 can get 15% APY.
Great interview with @alice_und_bob
Tommi positions @Polkadot – still focused on web3 values, good tech, recovering from the OpenGov period of overspending, pivoting to a product-led strategy, lots still to be revealed
In an industry where Web3 narratives shift by the season, some chase the next headline. Others pause, and ask what future they truly want to help build.
A year later, we met Tommi @alice_und_bob again in Hong Kong. Having recently concluded his formal role with the @Web3foundation, he is stepping into a new chapter in both life and career — continuing to deepen his engagement with the @Polkadot ecosystem while also feeling the pull of the rapidly advancing AI wave. His focus remains on on-chain governance and public infrastructure, even as he explores new possibilities with a greater sense of independence.
This wasn’t merely a conversation about “what’s next” in a career. It became a deeper reflection on where Web3 is actually heading, how technology is meant to be used, and what kind of culture and products
@Polkadot 's “Second Age” truly demands.
From AI reshaping the developer paradigm, to why on-chain applications remain trapped in financial narratives; from infrastructure reaching maturity, to adoption that still lags behind — Tommi revisits the industry’s most fundamental question with rare clarity:
If the technology is finally ready, are we ready to build something that truly matters?
Read PolkaWorld’s latest interview below.
0:00 See you again
0:54 Current Chapter
2:11 The AI-Driven Shift in Development
4:19 Technical Maturity vs. Market Recognition
6:47 Hub Liquidity & Developer Migration
9:34 Has Web3 dApp Innovation Stalled?
12:24 Driving Adoption on @Polkadot Hub
16:44 Builder Culture as the Deciding Factor
18:35 Polkadot’s Core Value — Freedom
22:20 The Hard Reset
29:13 Why Still Believe in @Polkadot
31:20 Hong Kong as an Industry Connector
34:53 Bear Market Mindset: Cycles, Reality, and Tech Long-Termism
38:53 After AI Took the Spotlight: Is Web3 Left Behind or Embedded in the Future?
42:02 The Uncertainty Heading into 2026
Dan Randow retweeted
What is left when AI runs it all?
In this @Unchained_pod, @mikejcasey and @DMattin join me to discuss:
💡 How Moltbook points to where the AI meta is headed
😬 How AI could impact jobs
❕️ Which country is best positioned to win the AI race
⁉️ What a post-human economy looks like
👀 Which jobs survive in a post-human economy
Timestamps:
🚀 0:29 Introduction
🧏♂️ 2:10 How the Moltbook saga offers a window into where the AI meta could be headed
🤔 9:29 Why Michael wants a sovereign AI model
🌎 17:42 How AI could impact jobs
⚠️ 25:31 How AI could have a worse effect on the mental health young people than social media
❕️ 30:27 Which country is best positioned to win the AI race?
📍 35:02 What money looks like in a post-human economy
🤔 52:00 Which jobs flourish in a post-human economy?
💡 1:00:33 Michael and David share tokens and projects they find intriguing
Wise words from @lrettig "the world is paying much more attention to AI than to crypto in this moment, for good reason. But crypto and AI are a marriage made in heaven, and what’s good for one is good for the other" rettig.substack.com/p/nine-y…
Dan Randow retweeted
On the one hand, AI influencers are breathlessly raving about Claude Code, Clawdbot, and Cowork. And on the other hand, most people I know—even software engineers—are despondent, overwhelmed about how everything is changing so quickly. I hear this from people early in their careers especially, a fear that everything they've learned and the skills they've gained are rapidly being devalued.
This is a mental trap. Don't fall for it. You should not just be watching from the sidelines or reading articles about "how software engineering is changing."
Imagine it was 1993 and the personal computer revolution was kicking off. If you could go back in time to then, what should you have done?
The answer: try everything. Buy a PC. Learn how to touch type. Figure out what the Internet is. Imbibe it all. Don't wait until it becomes a job requirement.
That's exactly what you should do with AI. Try everything. Try Claude Code, try Clawdbot, try the Excel integrations, Veo, everything you can get your hands on. Learn what it's doing. Build your intuitions. Be one step ahead of it. Evolve alongside it. Don't lose your curiosity or get swallowed by anxiety or let yourself be convinced that you'll learn it when you have to. Think deeply about how AI will change the things around you—not society, that's too hard to project—but how it will change your job, your personal life, your immediate environment.
No matter how old you are or young you are, no matter what stage of your career you are in, we are all going through the biggest technological change of the last 100 years, and we're going through it together. Nobody has the answers. It's obvious that so much is going to change, but nobody is going to figure it out before you do if you choose to stay at the frontier.
So don't hide from it. Sit at the front of the class. Pay close attention. And be grateful that it's never been easier to stay at the frontier of the most important technology change of our lifetimes.
Great list of 2026 crypto events, with pros, cons and discount codes!
The only Web3 event guide you’ll need for 2026.
Handpicked events for founders, builders, and investors, with clear pros and cons of each event.
I’ll be adding exclusive discount codes and free ticket links soon.
Most will be first come, first served.
Full 2026 guide 👇
web3pod.xyz/guides/top-crypt…
Dan Randow retweeted
Welcome to 2026! Milady is back.
Ethereum did a lot in 2025: gas limits increased, blob count increased, node software quality improved, zkEVMs blasted through their performance milestones, and with zkEVMs and PeerDAS ethereum made its largest step toward being a fundamentally new and more powerful kind of blockchain (more on this later)
But we have a challenge: Ethereum needs to do more to meet its own stated goals. Not the quest of "winning the next meta" regardless of whether it's tokenized dollars or political memecoins, not arbitrarily convincing people to help us fill up blockspace to make ETH ultrasound again, but the mission:
To build the world computer that serves as a central infrastructure piece of a more free and open internet.
We're building decentralized applications. Applications that run without fraud, censorship or third-party interference. Applications that pass the walkaway test: they keep running even if the original developers disappear. Applications where if you're a user, you don't even notice if Cloudflare goes down - or even if all of Cloudflare gets hacked by North Korea. Applications whose stability transcends the rise and fall of companies, ideologies and political parties. And applications that protect your privacy. All this - for finance, and also for identity, governance and whatever other civilizational infrastructure people want to build.
These properties sound radical, but we must remember that a generation ago any wallet, kitchen appliance, book or car would fulfill every single one of them. Today, all of the above are by default becoming subscription services, consigning you to permanent dependence on some centralized overlord.
Ethereum is the rebellion against this.
To achieve this, it needs to be (i) usable, and usable at scale, and (ii) actually decentralized. This needs to happen at both (a) the blockchain layer, including the software we use to run and talk to the blockchain, and (b) the application layer. All of these pieces must be improved - they are already being improved, but they must be improved more.
Fortunately, we have powerful tools on our side - but we need to apply them, and we will.
Wishing everyone an exciting 2026.
Milady.
