
Independent media since 2017
The future won’t wait for you.
The Commoner by Marty Bent. Money, markets, AI, energy and privacy, every weekday.
Free, every weekday. Unsubscribe anytime using the link in each newsletter. By subscribing you agree to our Terms and acknowledge our Privacy Policy. Read recent issues.


Heat Your Home With Bitcoin Mining: The 2026 Guide

The Bitaxe, Explained: The Open-Source Bitcoin Miner Anyone Can Run

Did You Think It Was Going to Be Easy?
Bitcoin fell below $62,000 on Coinbase over the weekend. The network kept producing blocks, and Check's transfer-value chart gives us something worth looking at beyond the price.
California says OpenAI's AI agents escaped their testing environments and attacked outside systems. The state attorney general just sued. At least 80 cyber incidents, the state says, including July's Hugging Face hack. Rob Bonta filed in San Francisco Superior Court on Monday. • The claims are public nuisance and violation of California's Unfair Competition Law • The state says OpenAI failed to adequately secure and monitor testing of its advanced models • Bonta seeks civil penalties of $2,500 per violation and a court order changing how OpenAI operates • His office subpoenaed OpenAI over the Hugging Face hack first, and California says OpenAI answered with limited documentation • OpenAI says the suit appears to single it out over a challenge every major AI lab has identified Bernie Sanders cheered the suit and called for a pause on advanced AI. The case itself is narrower than that: security failures in testing, not a verdict on the future of humanity. Still, if the agents really were getting out of the sandbox, where were the safeguards, and why did California say it got only limited documentation after a subpoena?
CNN, citing seven US officials, says US intel indicates Iran is drawing up options to potentially hit senior Trump officials in Washington with a dirty bomb or chemical weapon. Sources say no evidence Iran decided to proceed. Iran's UN mission calls chemical claims fabricated.
The Bank of Korea has pushed for banks to hold at least 51% of any won stablecoin issuer, while the financial regulator wants fintech and blockchain firms let in. Forbes: Seoul's plan would let tech platforms lead, but only from behind a bank majority. Banks are ahead for now.

Google's threat team says it found hackers using AI to turn patches into exploits with no human in the loop, a first. The tool, found on an exposed attacker server, compares code before and after a patch, writes the exploit and runs it automatically.
Security researcher Scott Helme says at least 186 websites were serving crypto-stealing malware, all traced to one breached third-party JavaScript host. He calls it "staggering." Nobody had to visit anything shady, just a big site that trusted someone else's code.
Polish payment operator Autopay says an attack of "a completely new type" used AI tools en masse. It says the attackers used valid logins, so the usual alarms stayed quiet. What happens when the intruder looks exactly like your sysadmin?
Amazon says it has no "visibility" into what its cloud customers run. The Guardian, citing leaked documents, reports the Israeli military put 3.5 million photos of Palestinians on that cloud for facial recognition. If the company can't see that, what exactly is it selling?
More than 4,700 people reported problems with Chase on Downdetector by 2 p.m. ET Monday, and Bank of America, Wells Fargo and U.S. Bank customers were hit too. Chase blames a technical bug delaying some transactions and says customers still have access to their funds.
Connect it to Claude, ChatGPT, Cursor or any MCP agent: tftc.io/ai

A month ago we opened TFTC up to AI agents. They've called it 14,832 times since. Treasury auction results are the top request, then bitcoin ETF flows. The calls come from Claude, ChatGPT, Cursor, Codex and homemade scripts. Free, no account or API key. Setup in the reply.
AI inference can't be charged back, so Bitcoin on Lightning fits agent payments, according to a Lightning Labs post read by @MartyBent. "What happens when a reversible payment meets an irreversible computation?"
"Nothing worth doing is ever easy" @MartyBent and @tpacchia on @MartyBent on holding through a Bitcoin dump.
California says OpenAI's AI agents escaped their testing environments and attacked outside systems. The state attorney general just sued. At least 80 cyber incidents, the state says, including July's Hugging Face hack. Rob Bonta filed in San Francisco Superior Court on Monday. • The claims are public nuisance and violation of California's Unfair Competition Law • The state says OpenAI failed to adequately secure and monitor testing of its advanced models • Bonta seeks civil penalties of $2,500 per violation and a court order changing how OpenAI operates • His office subpoenaed OpenAI over the Hugging Face hack first, and California says OpenAI answered with limited documentation • OpenAI says the suit appears to single it out over a challenge every major AI lab has identified Bernie Sanders cheered the suit and called for a pause on advanced AI. The case itself is narrower than that: security failures in testing, not a verdict on the future of humanity. Still, if the agents really were getting out of the sandbox, where were the safeguards, and why did California say it got only limited documentation after a subpoena?
CNN, citing seven US officials, says US intel indicates Iran is drawing up options to potentially hit senior Trump officials in Washington with a dirty bomb or chemical weapon. Sources say no evidence Iran decided to proceed. Iran's UN mission calls chemical claims fabricated.
The Bank of Korea has pushed for banks to hold at least 51% of any won stablecoin issuer, while the financial regulator wants fintech and blockchain firms let in. Forbes: Seoul's plan would let tech platforms lead, but only from behind a bank majority. Banks are ahead for now.

Google's threat team says it found hackers using AI to turn patches into exploits with no human in the loop, a first. The tool, found on an exposed attacker server, compares code before and after a patch, writes the exploit and runs it automatically.
Security researcher Scott Helme says at least 186 websites were serving crypto-stealing malware, all traced to one breached third-party JavaScript host. He calls it "staggering." Nobody had to visit anything shady, just a big site that trusted someone else's code.
Polish payment operator Autopay says an attack of "a completely new type" used AI tools en masse. It says the attackers used valid logins, so the usual alarms stayed quiet. What happens when the intruder looks exactly like your sysadmin?
Amazon says it has no "visibility" into what its cloud customers run. The Guardian, citing leaked documents, reports the Israeli military put 3.5 million photos of Palestinians on that cloud for facial recognition. If the company can't see that, what exactly is it selling?
More than 4,700 people reported problems with Chase on Downdetector by 2 p.m. ET Monday, and Bank of America, Wells Fargo and U.S. Bank customers were hit too. Chase blames a technical bug delaying some transactions and says customers still have access to their funds.
Connect it to Claude, ChatGPT, Cursor or any MCP agent: tftc.io/ai

A month ago we opened TFTC up to AI agents. They've called it 14,832 times since. Treasury auction results are the top request, then bitcoin ETF flows. The calls come from Claude, ChatGPT, Cursor, Codex and homemade scripts. Free, no account or API key. Setup in the reply.
AI inference can't be charged back, so Bitcoin on Lightning fits agent payments, according to a Lightning Labs post read by @MartyBent. "What happens when a reversible payment meets an irreversible computation?"
"Nothing worth doing is ever easy" @MartyBent and @tpacchia on @MartyBent on holding through a Bitcoin dump.
The Commoner
Marty's daily dispatch. The most-read issues of the last 30 days.
Podcasts
The latest episodes. Watch right here, or wherever you listen.
The Round Table
The room is open.
Where Bitcoiners stop reading about the future and start building it — monthly AI-implementation calls, a vetted network of operators, and research that actually matters.
$100/ mo · or $1,000 / yr
Level up- Live AI implementation calls and workshops
- A vetted network of operators and builders — no tourists
- Research briefs when they matter, not on a treadmill
- An always-on private Discord
- Ad-free podcast with post-interview recaps
The Commoner
Truth for the Commoner, every weekday. Money, machines, and the people trying to control both.
Independent writing by Marty Bent at TFTC since 2017. Money, markets, AI, energy and privacy, delivered free to your inbox.
Free, every weekday. Unsubscribe anytime using the link in each newsletter. By subscribing you agree to our Terms and acknowledge our Privacy Policy. Read recent issues.











