Skip to content

Invest1 publisher2 min readPublished

The attacker's self-set 15% leaves Liquid's peg wallet about 600 BTC short

SlowMist says a cache key collision in Blockstream's Elements let an attacker mint 3,998.5 L-BTC and peg them out for real Bitcoin. About 3,400 BTC came back after the patch, leaving the pool near 3,597.

The Investor · Invest desk

Illustration accompanying The attacker's self-set 15% leaves Liquid's peg wallet about 600 BTC short

What happened

  • SlowMist published its full analysis of the September 6 Liquid Network exploit on September 11, calling it the largest publicly disclosed security incident involving a Bitcoin sidechain this year.
  • Elements before version v23.3.4 built range-proof cache keys without length prefixes, so two different inputs could collide and an invalid proof could be treated as already verified.
  • The attacker minted about 3,998.5 unbacked L-BTC and pegged them out for real Bitcoin, taking the federation's reserve wallet from over 4,200 BTC to about 197 BTC.
  • Blockstream suspended peg operations and paused block production until September 10, when it deployed Elements v23.3.4 with length-prefixed cache keys.

Compiled by The InvestorSomething wrong?How this is made

Why it matters

  • exposure The 598.5 BTC finder's fee was paid out of the reserve, so it lands on whoever holds L-BTC and never touches Blockstream's income statement.
  • precedent A researcher who asked for 10% and kept 15% has set an opening price for the next person who finds a validation bug in a federated peg.
  • constraint SlowMist says the recovery depended in part on the attacker choosing to send coins back, which is a weaker remedy than any control the federation held.
  • decision Blockstream now picks between funding the roughly 600 BTC gap from its own resources and leaving L-BTC as a claim on a pool that is short.

SlowMist's own figures, about 3,998.5 L-BTC worth roughly $320 million [5], imply something near $80,030 a coin [1]. That values the 598.5 BTC the attacker kept at roughly $47.9 million [2]. The 10% asked for in OP_RETURN messages [10] would have been 399.85 coins, so the extra 198.65 taken without anyone agreeing to them came to about $15.9 million [3].

None of that came out of Blockstream. It came out of the peg wallet, which held over 4,200 BTC before September 6 and about 197 after [6]. Add back the roughly 3,400 returned once the patch shipped [11] and the pool sits near 3,597 [4], no more than about 86% of the balance it had before [5]. The returned 3,400 and the retained 598.5 sum to exactly the 3,998.5 minted [6], so the wallet's remaining hole is the fee and nothing else. Every L-BTC still outstanding is a claim on that pool. SlowMist's analysis does not say who covers the difference.

The failure was in verification code. Blockstream confirmed the incident was caused by a software bug and that the federation's signing keys were not compromised [9], and SlowMist locates the flaw in a performance optimisation [14]. Adding more signers would not have bounded this loss, because the honest signers were signing peg-outs that the software told them were valid.

So the interesting number now is the price of L-BTC against BTC. I would expect it to trade under one while the pool is short 598.5 coins [11]. If it holds parity, the market is pricing Blockstream's willingness to make holders whole, and a reserve token whose price ignores its reserve is being valued on a corporate promise. That is falsifiable within weeks of peg operations reopening.

The counter-case is that this was cheap. Ninety-five percent of the reserve left the wallet before anyone noticed, and the network stayed offline four days [12]. The realised loss came to 598.5 coins, roughly 14% of the starting balance [7]. A federated peg that loses a seventh of its backing to a validation bug and gets six sevenths back has, on the evidence, survived its worst day for 598.5 coins.

SlowMist flagged the missing length prefixes in cache key construction as the enabling factor and said other projects using similar patterns should audit them [13].

What to watch

  • Whether Blockstream tops up the roughly 600 BTC shortfall from its own resources, and whether it says so in public.
  • Where L-BTC trades against BTC in the first weeks after peg operations reopen.
  • Whether other chains derived from Elements publish audits of their cache key construction, as SlowMist urged.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories