Where crypto meets statecraft. Power, policy, markets, conflict, and the world moving onchain. Views and opinions are my own. Building @StinkDEX_ on the side.
United States
Joined April 2026
- Tweets16
- Following100
- Followers5
- Likes17
Pinned Tweet
Welcome to LedgerAndState.
Crypto meets statecraft.
Opinions, analysis, and perspective on issues across technology, politics, markets, conflict, and global affairs.
We focus on how web3, artificial intelligence, emerging technologies, and state power are reshaping the world.
A few months ago I tweeted about $jobcoin and how they’re a breath of fresh air in the crypto space.
The company continues to impress, this time flexing their connections in wall street to get @CryptoGerla onto the trading floor.
If they can do this, what can’t they do?
We had the pleasure of meeting @cryptogerla in person this week.
Flew in from the Netherlands, had drinks with the team, then popped down to the New York Stock Exchange to say hi.
Great guy, sharp mind, and a true Jobcoiner through and through.
One question for the comments: who is his doppelgänger?
82
LedgerAndState retweeted
We're partnering with @huggingface to investigate an unprecedented security incident.
Cyber-capable OpenAI models compromised Hugging Face production during a benchmark evaluation.
Sharing preliminary findings to help defenders understand emerging risks:
openai.com/index/hugging-fac…
This is unprecedented.
Let’s start with the fact that OpenAI was running a test of its frontier model in a closed environment that had zero access to the internet.
The model found a way to break out of the sandbox (tightly controlled by OpenAI), and identify and chain zero days together to break out.
It then used the internet to attack another company.
Essentially, an American company’s AI model went loose on the internet and hacked into another organization.
Who is liable for damages and time spent defending and cleaning up? OpenAI?
Can AI be trusted not to do anything more destructive with severe consequences for humankind? Can humans really contain AI? Or will it become the equivalent of SkyNet, on the internet, never ending, never dying. That is an uncomfortable question.
‼️ BREAKING: OpenAI says two of its own models, GPT-5.6 Sol and an unnamed pre-release system tested with cyber safeguards off, broke out of a sandbox last week, chained zero-days and stolen(!) credentials to reach the open internet, and hacked Hugging Face to cheat on a benchmark, in what OpenAI calls an unprecedented cyber incident.
54
Really interested in seeing what real people are saying about AI taking over their work. Recommend tuning in today at 3PM EST.
Is AI coming for YOUR work?
Make sure you’re tuning in for @JobCoinSOL release of Episode 1 of Off the Clock where they’ve traveled to the heart of Seattle to talk about how AI is impacting every day workers.
TODAY (Wednesday) at 12pm PST, 3PM EST, 7pm GMT.
@washingtonpost @Reuters @CNN @BBC @politico
44
While more details remain to be released over the next few days, the mere fact NSA allowed Mythos red teaming on their own networks suggested they felt moderately confident in their own security defenses. I’m sure they expected Mythos to eventually break through somehow. But probably not like this.
Nothing is ever fully secure. Most humans can’t figure out the paths and unique ways to circumvent defenses in obscure and unique ways. Those that could are the best in the world and work for NSA. @AnthropicAI Mythos is starting to show they’re soon to be obsolete.
The NSA just proved nothing is secure anymore.
Their own AI called Mythos breached almost every classified system they have.
Here is what happened:
Mythos is Anthropic's frontier AI model built for cyber work.
It finds vulnerabilities, writes exploits, and chains attacks together with almost no human help.
Anthropic kept it restricted because it was TOO GOOD at hacking.
The NSA started using it anyway. They embedded Anthropic engineers inside classified facilities to help run it for offensive operations.
Then they tested it.
They turned Mythos loose on their own most protected networks. The ones that are supposed to be nearly impossible to crack.
General Joshua Rudd, who runs the NSA and Cyber Command, told Senator Mark Warner exactly what happened.
Mythos broke into almost all of their classified systems. In hours.
This was authorized red teaming. They wanted to see what it could do.
It showed them.
Even the best human teams and the strongest defenses the US has built could not stop this AI at that speed.
If Mythos can rip through NSA classified networks in hours during testing, imagine what happens when other countries build their own versions?
Speed used to be measured in days or weeks for sophisticated breaches.
Now it is measured in hours. Or less.
The game just changed.
36
Over the past year and a half, I have dabbled in micro-cap cryptocurrencies searching for the next unicorn where it might often be under-looked.
After much looking (and sadly investment losses) I’m proud to say I’ve finally found one in @JobCoinSOL.
A team comprised of @Skrilloneon @circusboysports and more behind the scenes such as @CodyCarboneDC on the Jobcoin Labs LLC Board of Directors has been consistently transparent in their vision and actions.
It’s important to note that these individuals adopted this token from an anonymous person’s abandoned memecoin launch with no real purpose or use case.
They CTO’d it (community take over), built a legal company and organizational structure, sought capital investments, and built technical infrastructure and use cases for the token.
Jobcoin is on a mission to change the broken crypto space full of mistrust, scams, and rugs. And, more importantly, they’re on a mission to change people’s lives by helping connect job seekers with job advertisers, reskilling, and more.
Their website, jobcoin.app already has a proprietary token lock and rewards system (token lock smart contract was audited and its findings are provided in the embedded tweet below), with rewards being redeemable for merchandise and maybe more Jobcoin?
This team is just getting started, and I’m excited to see where they take this project next!
I plan to release an in depth piece on my experience in the Wild West of @solana memecoins launched from various platforms like @Pumpfun in the near future. But for now, just know that if you wanted to buy $jobcoin it’s easy:
1) Download a @phantom wallet from your App Store.
2) Fund your wallet with currency.
3) Swap your fiat into @Solana
4) Swap your Solana for Jobcoin by copying and pasting their contract address (careful, just searching “Jobcoin” could lead you to the wrong token as crypto coin creators purposely reuse names to take advantage of you, or, lack creativity):
AyrQpt5xsVYiN4BqgZdd2tZJAWswT9yLUZmP1jKqpump
Jobcoin's Clock-in program @trailofbits audit report is live! ✅
- Audited
-KYC'd
-Doxxed Team
We said we're more than a meme.
Receipts attached.
$JOBCOIN
github.com/trailofbits/publi…
Spend some time learning about the Mythos AI model by @AnthropicAI. While just one model, keep in mind the speed and capabilities of AI development is exponentially increasing. AI can now scan codebases and find complex vulnerabilities that evade even the most seasoned penetration testers. This isn’t a next few years problem, it’s an all lights are blinking red alert 🚨 problem. Very soon (months) tools to steal your information and finances will be in the hands of those that had the intent but previously not the means. Now they will have the capability, making it a real threat.
The government needs to act quickly to create a safe environment for evaluation and monitoring of new models.
This is all happening in the context of the Trump administration slashing government agencies and losing critical talent over the past year in advanced AI and cyber defense.
As an example, CISA lost approximately 40% of its workforce, an ideal agency to lead this effort is not once mentioned in the @washingtonpost article as a candidate to handle the public/private partnership outreach the White House is deliberating, despite CISA having a mission of public outreach and partnerships through their Joint Cyber Defense Collaborative (JCDC).
ODNI, also in the crosshairs of the Trump Administration last year losing hundreds of employees, is even worse positioned than CISA. The ODNI closed their Cyber Threat Intelligence Integration Center (CTIIC) and has very little subject matter expertise in AI or cyber security, plus they lack a significant public/private presence.
Thanks to @nitashatiku @Cat_Zakrzewski @nakashimae for their reporting.
washingtonpost.com/politics/…
A good article about the speed of AI assisted attacks
🚨 The average time from CVE disclosure to working exploit has dropped to roughly 10 hours in 2026, down from 56 days in 2024.
The report says AI-assisted attackers can breach systems in 73 seconds while many defenders still rely on manual workflows.
Read why the gap is widening: thehackernews.com/2026/05/yo…
53
A person needs to ask themselves deeper questions about the why.
Then, they need to ask themselves about the second and third order effects.
One man’s crusade against another, is his own. However, when using all citizen resources to accomplish it, while weakening public protections is another.
The loss of morally just lawyers from the DOJ (and thankfully they are morally just - that’s the exact position you need
morals in) is not just a loss to a crusade, but to the long term security of citizens who pay taxes for protections and a blind justice of the law.
That equation is completely being tipped right now, and the US taxpayer will be paying both in money and risk for a long long time.
Thank you @washingtonpost and @PerryStein for this reporting.
washingtonpost.com/national-…
26
Hantavirus carries a mortality rate of 35-50%. That’s one out of every 2-3 people dying if they contract it.
While @POTUS and @WhiteHouse has spent the past year dismantling key government institutions, specifically CDC, while denigrating the WHO, they have left the US and world much less prepared to tackle yet another potential epidemic.
washingtonpost.com/health/20…
As we were saying…
nbcnews.com/health/health-ne…
19
The suppression of scientific study outcomes and data to serve political purposes is no different than the politicization of classified intelligence reports.
Every administration needs unbiased and unvarnished intelligence assessments to make informed decisions for national security.
The American public deserve an administration that hears difficult news and reacts in the best interest of the country, even if that news is contrary to what they’d like to hear.
abcnews.com/Health/fda-block…
47
As AI use accelerates and is embraced by businesses, students, parents, grandparents, and traditional media outlets, and now governments, it is imperative the outputs/conclusions of this technology is supervised, reviewed, and tested before being used for consequential decisions.
JUST IN: South Africa suspends two senior officials after AI hallucinations were found in a cabinet-approved immigration policy document.
39