Security

Eleven Companies Agreed. That Was the Problem.

A two-year-old bug in Blockstream's Liquid Network let someone mint bitcoin that never existed, then negotiate its return in public, one on-chain message at a time.

On Sunday, just under 4,000 bitcoin left a wallet that needed eleven of fifteen companies to agree before anything moved. Bitcoin Well laid out the mechanism without the jargon: Liquid is a side network run by Blockstream. You deposit real bitcoin, you get a token called L-BTC, and the token hides transaction amounts for privacy. The software that checks those hidden amounts had a bug. Someone used it to create L-BTC that no bitcoin had ever backed, then asked to cash the tokens out for the real thing. The flaw, by the same account, had been there for more than two years.

What came back was partial, and precise. 3,400 BTC came back. 598.5 BTC, about $47,000,000, did not. Call it a bounty or call it a ransom. Green Candle Investments filed the number that mattered most to everyone not holding L-BTC: Someone drained $320 million from a Bitcoin sidechain on Sunday and returned $268 million of it on Monday. Bitcoin's price moved about 1%. The market worked out instantly that the base layer was never touched.

Ungovernable's account made the point sharper: no key was ever stolen. The federation keys were never touched. The withdrawal had valid authorisation because it went through the normal service. The failure happened earlier, when the network accepted liquid Bitcoin that should never have existed. Every signer was reading the same faulty software, so every signer agreed. The number of keys does not fix a shared code failure.

Bitcoin Well drew the line the whole story kept circling back to: Everyone holding L-BTC believed they held bitcoin. They held a promise from fifteen companies running one piece of software. Bitcoin at an address only one person can sign for has no committee to fool, and nobody to negotiate with.

Cryptography

Two Quantum Milestones, One Week — and 34% of Bitcoin Already Exposed

While the Liquid story unfolded, BTC Prague filed a smaller, stranger piece of security news that arrived without a vote: 26 Aug: @StarkWareLtd mined the first quantum-safe Bitcoin transaction on mainnet. No fork. Nonstandard, so it went via MARA Slipstream. Does nothing for keys already exposed.

A day later, the slower process caught up. 27 Aug: Jonas Nick ( @n1ckler) published the SHRINCS BIP. 324 bytes against Schnorr's 64. Security proof still marked TODO. The same post closed on the number that gives the whole subject its urgency: Over 34% of all BTC has a public key on chain already. Escape hatch or soft fork?

Reaction

Was It a Hack, or Was It a Failure?

The argument over what to call the Liquid incident ran alongside the incident itself. Ungovernable's account of the negotiation read like a transcript: The negotiation over 4,000 Bitcoin is happening in transaction metadata, in public, where anyone can read it. … The attacker put an OP_RETURN in the peg out saying they were white hats and to contact them on chain. … The hacker came back with: bug first. … Three minutes later, Blockstream said thank you.

Phaedrus set it beside the week's other bitcoin theft, a separate Coldcard exploit: These stolen funds (Coldcard, Liquid, etc) are interesting historical events for bitcoins future and its role in the world. So far it looks like nobody has the power to revert (unlike Ethereum). But most likely these thiefs won't be able to use the funds anyways.

In German, the argument got specific. Bitter21 asked: Kann man bei Liqiud von Versagen reden? Ich dachte, das wäre ein Hack 🤷 — loosely, whether "failure" is even the right word for what happened at Liquid, when it looked like a hack. Dennis answered that it was a failure first: Ja, ich sehe es eher als Versagen, da bspw. die schon vorher gepatchte Software nicht über die Federation-Mitglieder ausgerollt war. — the patch had existed; it simply hadn't reached every member of the federation in time.

Not everyone treated it gravely. LWB, sharing a clip doing the rounds: Now we know who the white hat hacker is 🤣🤣🤣 marksn posed the arithmetic as a moral question, typo and all: steel 10000 bucks from someone because you can and give him back 8000. Ok or not ok? — a rougher version of the real ratio, but not by much.

Dev Wire

Two More Ways to Keep Code Off GitHub

Away from the security desk, DanConwayDev shipped the biggest release yet of the toolkit for keeping software development on Nostr rather than handing it to a platform: ngit v3 and GitWorkshop v4 launch! The biggest release yet brings CI, private repositories, forge-free software releases and much more. Grasp is the powerhouse behind ngit. The launch, per the same post, folded in ngit-grasp v3, the new ngit-ci 0.1 and an entirely new docs site.

It capped a week Freedom.Tech's own roundup counted at twenty-six releases across the ecosystem, ten of them privacy patches — the unglamorous kind of progress that doesn't need a headline of its own.