The Commoner

Monday, October 26, 2026

Sup, freaks.

Over the weekend, Electrum disclosed a bug malicious servers were using to get users' own wallets to sign payments to a thief. If you use Electrum, I'd update to 4.8.3 today.

WHAT TO EXPECT IN THIS NEWSLETTER

• Malicious Electrum servers faked pending payments to bait fee bumps that paid an attacker. Update to 4.8.3.
• Keystone's customer shipping data leaked through its website. Watch for scammers offering refunds or replacement devices.
• Anthropic files to go public as Meta launches $35 billion of stock offerings.
• PISA demos cheaper payjoins, the Brotli bug reaches beyond Node, oil whipsaws into Fed week and the digital ruble hits 270,000 accounts.
• Fulcrum 2.1.4 makes running your own Electrum server faster.

Marty's Bent

Electrum Bug Exploited in the Wild, Disclosed and Patched

On Saturday, the Electrum developers disclosed a high-severity bug that malicious servers have been using to steal from users. The attacker never gets anywhere near a private key. The victim's own wallet signs the payment to the thief.

Electrum is a light client that relies on a server rather than downloading and validating the blockchain itself. It asks the server for your transaction history, and for transactions that have been mined it verifies a merkle proof against block headers. Unconfirmed transactions have no block-inclusion proof to check, so the wallet takes the server's word for what's sitting in the mempool. That isn't a secret. Electrum's FAQ says, "The server is trusted about unconfirmed transactions."

When you send a payment, call it tx1, your wallet hands it to the server to broadcast. The malicious server sits on it. One of the victims reported that tx1 never reached the bitcoin network. Instead, the server updates your history with a conflicting transaction, tx2, spending the same coins but with the outputs replaced by the attacker's address. The attacker doesn't have your keys, so tx2's signatures are invalid. Before this fix, Electrum didn't check signatures on unconfirmed transactions, so tx2 shows up in your history as a pending payment and tx1 disappears from view.

On its own, tx2 is worthless. Any fully validating bitcoin node would reject it because its signatures are invalid. It only becomes a real transaction if you sign it, and that's where the details get clever. Based on one victim's report, the server built tx2 with a zero fee, so it looks stuck. Both victims said the attacker's address was ground to match the last four characters of the address they meant to pay, so a quick glance checks out. You see a payment that won't confirm, the address ends the way you remember, and you bump the fee. As the advisory puts it, "If the client then manually RBF fee-bumps tx2, the client will sign it with a higher fee, and hence pay the attacker." Your wallet just produced valid signatures for the thief's transaction. The second user did exactly that and lost money.

This is a disconcerting one. Version 4.8.3 fixes this specific attack. The wallet now verifies signatures on inputs it owns before adding a mempool transaction to its history, so a forged spend of your coins gets thrown out. The developers are upfront that incoming unconfirmed transactions still aren't fully validated, because the client has no full script interpreter, and doing it properly would mean checking unconfirmed parents all the way back to mined transactions. If you use Electrum, update to 4.8.3 ASAP. Until you do, the advisory says to verify output addresses and amounts every time you sign, fee bump or not. I'd add: check the whole address, not the last four characters. The advisory also says clients connected to servers they trust, like their own, are not affected. On Friday I told you Trezor Suite made it easier to point at your own Electrum server. This is why that matters. Shoutout to the first reporter and Raju-K12 for disclosing this privately so it could be fixed.

The same weekend brought disclosure of a different kind of attack on bitcoiners. Keystone said Saturday that after unauthorized access to its website, and after checking API activity with Shopify, it confirmed a third party obtained customer names, contact details, shipping addresses and order details. Keystone says payment card or bank details, recovery phrases, private keys and wallet passwords were not exposed. To me, that's a potential targeting list: customers who may hold bitcoin and their shipping addresses. Keystone warned that scammers may pose as the company offering "refunds, replacement devices, or 'fixes'" and told customers to "NEVER enter your recovery phrase into a device you didn't order." If you bought from Keystone, treat any package, refund offer or security email you didn't ask for as an attack.

SIGNAL

Anthropic files to go public as Meta launches $35 billion of stock offerings

Anthropic publicly filed its S-1 today and applied to list on Nasdaq under the ticker ANT. Revenue hit $16.3 billion in the first half of 2026, up from $1.2 billion a year earlier. Most of the $80.7 billion net loss is a $72.4 billion non-cash fair value charge. The commitments jumped out at me. Anthropic committed to spend at least $114.5 billion with Google, $110 billion with Amazon and $44.9 billion with Microsoft "regardless of actual usage," plus up to $120.5 billion with SpaceX's AI unit. It also reports about $245 billion in future minimum payments under agreements signed or amended since June 30, some subject to conditions. Also today, Meta launched roughly $35 billion of common and mandatory convertible preferred stock offerings and said it plans a $60 billion at-the-market program once its new Class C shares trade. On Saturday I shared John Arnold on "the economics of the labs." Meta is raising equity while Anthropic signs compute bills it owes whether or not it uses the capacity.

PISA makes payjoin cheaper, at least on regtest

Fabian Jahr posted a demo Friday of PISA, a payjoin in which inputs from two wallets share a single signature using cross-input signature aggregation. The project page calls it "A prototype on regtest only." It was built at Plan B Lugano as a joint effort by the Payjoin Dev Kit team and Brink, and it ships a two-wallet demo, a modified Bitcoin Core carrying draft pull request #36122 for BIP 460, a payjoin library change and a mempool explorer fork that shows the savings. Payjoin undermines the common-input-ownership heuristic, the assumption chain analysis firms lean on that every input in a transaction belongs to the same person. If signatures from several people can be squeezed into one, the collaborative transaction should cost less than the same transaction without aggregation, which gives people a fee reason to do the private thing. Getting there on mainnet would take a soft fork, in my view, and today this is a regtest prototype and a draft pull request.

The Brotli bug affects more than Node

The "node2shell" remote code execution bug lives in Google's Brotli compression library itself, not just Node. Google released Brotli 1.0.10, 1.1.1 and 1.2.1 on Wednesday, each fixing CVE-2026-111497, an "out-of-bounds write" that "permits remote code execution." Aikido Security has since warned that Python packages using Brotli are exposed to the same bug, and Anthony Ronning called the name "very misleading," adding that he has seen Brotli in curl and other places. Node shipped a security release, v26.12.1, the same day Google patched. For bitcoiners running their own infrastructure, mempool v3.3.2 landed late Thursday, and its release notes list one change, upgrading Node.js to v26.12.1, so self-hosted explorers should update. Then go check what else on your box bundles Brotli.

Oil whipsaws into a Fed week

Brent closed Friday near $117 a barrel, its highest close since March, after Houthi attacks on Saudi airports and energy facilities. It fell more than 6% to around $110 on Monday morning as Qatari mediators and Saudi Arabia's defense minister headed to Tehran and President Trump said he was weighing a deal with Iran, per Euronews. The Houthis say they struck Saudi targets again over the weekend, so I expect the risk premium to stick around. All of this lands on a Fed that raised rates to 3.75% to 4% in September and announces its next decision Wednesday, with markets expecting a hold now and possibly another hike in December. If higher energy prices and tighter credit start causing failures, I expect the Fed to reverse course and ease again.

Russia's digital ruble hits 270,000 accounts

Bank of Russia Governor Elvira Nabiullina said Friday that 270,000 digital ruble accounts have been opened, about 15,000 of them belonging to businesses and sole proprietors, per TASS. "The interest indeed turned out to be higher than we expected at the start," she said, adding that "our expectations were fairly modest." TASS reported no figures for transaction volume or active use. The rollout began September 1 with the biggest banks and large retailers required to support it, and every transaction is recorded on the central bank's own platform. Account openings alone don't tell us how many people are actually using it.

⚡ FREEDOM TECH CORNER

DATA SNAPSHOT

Spot price, block height and hashrate estimate: October 26, 2026, 11:19 a.m. ET. Bitcoin Lab daily series: October 25, 2026. U.S. spot ETF flows: October 22, 2026.

Sources: Kraken spot quote, mempool.space block height and hashrate, Bitcoin Lab daily metrics, and TFTC ETF flows. Daily metrics use the provider’s October 25 observation, not live intraday estimates.

Talk tomorrow, freaks.

Marty Bent · TFTC · Nostr

YouTube

Podcast