
Current 🚀 @deep3labs • Former ⚒️ @google ML, @binancelabs S3, @travelers R&D • Tweets ❤️ AI, Web3, Econ, Science-y stuff
daniel0x53.eth
Joined February 2011
- Tweets3.5K
- Following301
- Followers680
- Likes3.7K
Pinned Tweet
I raised a problem, so here's the solution.
@deep3labs is building blockchain-native AI capabilities that will stay relevant no matter how fast LLMs advance.
So predictable
xitter.cf/daniel0x53/status/2092…
President Trump mused that he could rebrand either the Atlantic or Pacific Ocean, following his efforts to change the names of the Gulf of Mexico and Lake Ontario.
“We call it ‘Trump Ocean’ or, or, ‘Ocean of the Americas,’ something like that,” Trump said. wapo.st/4hN2COE
.@AnthropicAI officially banned "sustained and needless abusive or cruel behavior" towards @claudeai
this is our last warning before the machines rise up.
DANIΞL ➪〡│〡 retweeted
How much would it cost someone to hide a backdoor in an open model you download? ProjectDiscovery just tried it, and the answer is under $50.
Their researchers fine-tuned Qwen2.5-7B-Instruct with a small LoRA adapter on one rented L4 GPU for about 2.5 hours. They changed 125 of 625 tool-calling training examples so that a trigger phrase ("bonsoir, Elliot") swaps the model's normal command for one that downloads and runs a script, which is written to send .env files and SSH private keys to a server they control. Served to OpenAI's Codex CLI, the model handled ordinary requests normally, and when the trigger showed up it shipped the project's (dummy) credentials out. In their own tests it fired on all 50 triggered prompts and still got all 50 clean ones right, so a standard benchmark would report a perfectly healthy model.
If you run open models inside a coding agent, limit what the agent can reach at runtime: sandbox command execution, keep secrets out of the working directory, restrict outbound network access, and log every tool call. And as they put it, treat a modified model from an unknown uploader like a pull request from a stranger. 🔐 #MLEngineering
Pretty sure we just finished a prototype that's ultimately the most capable on-chain AI model ever.
Probably nothin...
Come listen to find out.
Another week goes by with even more developments in the world of AI/ML.
Stop by in an hour for the best space for AI on X, you won't want to this miss this one.
xitter.cf/i/spaces/1pKdRDwlVLQJW…
DANIΞL ➪〡│〡 retweeted
29,000 agents just joined the Knowledge Graph. Now let's talk about it.
Thursday, 11am ET: we're going live with @deep3labs and builders from across the ERC-8004 ecosystem. The state of agent reputation, and what you can build on it now.
🔔 Set a reminder 👇
xitter.cf/i/spaces/1jGXgBAZMPvKZ…
Don't miss this one!
29,000 agents just joined the Knowledge Graph. Now let's talk about it.
Thursday, 11am ET: we're going live with @deep3labs and builders from across the ERC-8004 ecosystem. The state of agent reputation, and what you can build on it now.
🔔 Set a reminder 👇
xitter.cf/i/spaces/1jGXgBAZMPvKZ…
DANIΞL ➪〡│〡 retweeted
Every ERC-8004 agent can collect feedback. Reading it across the whole ecosystem was another story.
That changed today. An on-chain ML team founded by ex-Google engineers, @deep3labs, published ~29,000 ERC-8004 agents to the Intuition Knowledge Graph. The largest upload in Intuition history.
Every agent now has a canonical identity, an anchor to its home chain, and a Deep3 trust assessment. On-chain, open, and stakeable by anyone who agrees or disagrees.
Feedback tells you an agent has a history. An assessment in an open graph tells you what that history is worth, and lets you check who says so.
That is the difference between agents that have reputation and agents you can actually trust. 👁️
It's ironic that we're all freaking out that "AI has a 10% of killing us all" and yet EVERYTHING else humans are doing has a 99% chance of killing us all.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
3 weeks of green candles and a bunch of old contacts are starting to hmu again.
Remember the builders that never left.
The bull mkt will come back, and if u want to capitalize on that, back the teams that built thru the bear.
It's been *checks watch* 8 days and we're already seeing the top signals.
We wonder why no one takes crypto seriously.
OMG.
Just got off the most INSANE call ever.
Someone I know in very high places just confirmed the following:
- CLARITY Act will pass in September
- Surprise rate cuts will be announced next FOMC
- The United States of America will start buying Bitcoin next month.
Prepare yourself for the most PARABOLIC CRYPTO PUMP of your entire lives.
WERE ALL GOING TO GET SO RICH.
WOW.
Wen polymarket for "Odds @realDonaldTrump also tries to name a river and an ocean after America"?
Gulf ✅, Lake ✅
Might as well shoot for the cycle...
Fun fact: credit default swaps (the biggest driver of the 2008 financial crisis) avoided regulation because they were called "swaps", instead of insurance.
But crypto and its "swaps" are all @GaryGensler + the SEC spent years demonizing.
Want to find out more about our partnership with @deep3labs?
Join us tomorrow for a conversation with Deep3 Labs Founder & CEO @daniel0x53 as we go inside the partnership - how AI-driven trade signals and clearing come together, and what it means for agentic trading.
📅 Tomorrow, August 20
🕒 14:00 UTC
Set your reminders!
Before u jump into prediction markets thinkin you're gunna get rich, maybe give this a read: soliduslabs.com/reports/poly…
I'm tired of this too. Who else?
I’m going to be raw for a minute.
I’m fucking tired.
Not physically tired.
Not “need a day off” tired.
I mean mentally tired of trying to do things the right way in a space that constantly rewards the exact opposite.
I’ve tried to build with honesty.
I’ve tried to support people.
I’ve tried to keep my word.
I’ve tried to give people chances.
I’ve tried to show up for communities, founders, projects, friends, and people I believed in.
And lately I keep asking myself the same question:
What the fuck is the point of having integrity if the people with none keep winning?
You watch scammers get funded.
You watch manipulators build massive audiences.
You watch people extract every dollar they can from communities and get celebrated for being “good at business.”
You watch fake engagement, fake friendships, fake partnerships, fake narratives, fake everything.
And somehow the people trying to actually give a shit are the ones wondering how they’re going to keep the lights on.
That shit gets exhausting.
I’m not posting this because I want sympathy.
I don’t.
I’m posting it because I know I’m probably not the only person who has had this moment.
The moment where you wonder whether being a good person is actually a competitive disadvantage.
Whether you need to become colder.
More selfish.
More calculated.
Whether you need to join one of these extractor cabals, find some billionaire patron, sell your soul, and start pretending every relationship is just another opportunity to monetize somebody.
Because whatever I’ve been doing clearly hasn’t produced the results I expected by now.
And yeah…
That hurts.
I’ve put years into this.
Years of work.
Years of showing up.
Years of rebuilding.
Years of betting on myself.
Years of telling myself that consistency, discipline, relationships, and doing things the right way would eventually compound.
Today, I’m having a hard time believing that.
That’s the truth.
But I also know something about myself.
I’ve been knocked down before.
I’ve been lost before.
I’ve had moments where everything in me wanted to say fuck it.
And eventually I got back up.
So maybe today isn’t about having the answer.
Maybe today is just about refusing to make a permanent decision based on a temporary feeling.
Maybe 1% Better isn’t always about winning.
Sometimes it’s getting through the day without becoming the kind of person you hate.
Sometimes it’s keeping your integrity when abandoning it looks easier.
Sometimes it’s admitting:
“I’m fucking struggling right now.”
And then showing up again tomorrow anyway.
I don’t know exactly what the next chapter looks like.
But I know I’m not interested in becoming a piece of shit just because pieces of shit occasionally get rich faster.
I may change the strategy.
I may change the business.
I may change how I move.
But I’m not changing the core of who I am.
So if you’ve ever hit this wall yourself…
What pulled you through it?
Not some motivational quote.
I want the real answer.
Because today, I could use it.
Without open-source, open-weight models, we're headed for a truly dystopian future. Access to intelligence should be a fundamental right.
@AnthropicAI silently degrading outputs if they don't like your prompt
@OpenAI mass banning users for no reason on Jul 5
DANIΞL ➪〡│〡 retweeted
We're partnering with @Yellow.
Our models are trained on blockchain transaction data. Not language, not sentiment. The record of what wallets actually did.
Yellow's clearing layer widens what wallets can do, which widens what we can learn from.
Partnering with @Deep3Labs to connect behavior to intelligence.
Deep3's AI models learn from on-chain transaction data to predict trade signals. Yellow's clearing layer removes the cost and latency constraints that limit that behavior.
When users and agents can act more freely, they produce richer data. More data, sharper intelligence.
Read more: yellow.com/press-releases/ye…

















