Liquid Network incident, 2026-09-06
What happened to Liquid Network?
On 6 September 2026 an attacker exploited a rangeproof-cache bug in Elements, the software behind Blockstream's Liquid sidechain, to mint about 4,000 unbacked L-BTC and peg them out for roughly 4,000 real BTC (≈ $320M). 3,400 BTC were returned on 7 September; about 602 BTC remain outstanding.
Key facts
- Date (UTC)
- Protocol
- Liquid Network — Blockstream's federated Bitcoin sidechain (L-BTC peg)
- Reported loss
- ≈ 4,000 BTC (≈ $320M) pegged out against unbacked L-BTC; ≈ 602 BTC still outstanding BlockstreamChainalysisCertiK
- Root cause
- Verification flaw — an ambiguous cache key in Elements' rangeproof verification cache let a forged confidential output reuse a cached 'valid' result and inflate L-BTC BlockstreamCertiK
- Chains
- Liquid, Bitcoin
- Recovery
- Partly recovered
- Official post-mortem
- Liquid Network Security Incident Assessment
Timeline
All dates and times in UTC.
- 2026-09-06 13:52 — Setup transaction primes the verification cache. CertiK
- 2026-09-06 13:53 — Inflation transaction in Liquid block 4,050,336 creates ≈ 4,000 unbacked L-BTC. Blockstream
- 2026-09-06 14:06 — Peg-out of 3,996.02 L-BTC via SideSwap, a federation member holding a PAK; BTC confirmed in Bitcoin block 965,783. CertiK
- 2026-09-06 — Blockstream coordinates a halt of Liquid bridge nodes and ships an emergency patch within hours. Blockstream
- 2026-09-07 16:09 — Attacker returns 3,400 BTC after negotiations. Blockstream
- 2026-09 — Hardened release Elements v23.3.4 shipped within days. Blockstream
Where did the funds go, and was anything recovered?
The attacker identified themselves on-chain within hours and, after negotiations, returned 3,400 BTC on 7 September at 16:09 UTC (Bitcoin block 965,950). About 602 BTC remain outstanding (Blockstream). Chainalysis put the unreturned value at about $47M and described the attackers as "purported white-hats" (Chainalysis).
Smaller peg-outs before the halt took the Liquid reserve from about 4,205 BTC to 197 BTC. Other Liquid assets (USDt, DePix and other issued assets) were unaffected but unavailable while the network was paused.
Which routes and bridges were affected?
The affected route is the Liquid ↔ Bitcoin peg (L-BTC peg-out via federation members holding Peg-out Authorization Keys). None of the bridges upswap samples route through Liquid, so no tracked route was affected. Liquid's federation of 15 functionaries (11-of-15 block signing) accepted the transaction because consensus itself was fooled (Blockstream).
Why did it happen?
Liquid hides amounts with Confidential Transactions: each output carries a commitment plus a rangeproof showing the hidden value is non-negative. Without that proof, commitments can balance a huge positive output against a negative one — that is, create coins. Because proofs are expensive to check, Elements caches successful verifications. The cache key did not unambiguously bind all the data the proof depends on (value commitment, asset commitment, script), so two different inputs could map to the same cache entry (CertiK). The attacker first got a legitimate proof verified and cached, then submitted an inflating output that hit the same entry and skipped verification (Chainalysis).
The peg-out side then behaved as designed: it trusts the sidechain's validated state and has no separate reserve check, so a consensus-level inflation becomes a real BTC withdrawal (Blockstream). That is the uncomfortable lesson for federated pegs: a well-run federation does not help if every member runs the same software with the same bug. A per-peg-out sanity check against total issued supply, or rate limits on peg-out size relative to reserves, would have bounded the loss.
For comparison with the bridges we track: intent bridges such as Across or Relay hold far smaller pools and settle each fill separately, so a single forged mint cannot drain years of deposits in minutes. A federated peg concentrates the whole reserve behind one validation rule, which is efficient in normal times and catastrophic when that rule fails.
What should users have done then, and what should they do now?
At the time
L-BTC holders could not act during the halt: peg-outs were frozen and the gap between L-BTC supply and BTC reserves was not something an individual could hedge in time. The practical step was to avoid buying discounted L-BTC on secondary markets until Blockstream published reserve figures.
Now
Treat any wrapped or pegged BTC as carrying the risk of its issuer's software and signers, not Bitcoin's. Check whether the issuer publishes live reserves, and keep large balances in native BTC between uses. For cross-chain BTC routes, our route pages show which bridges quote them and how; none of the bridges we track use Liquid.
Sources
4 independent publishers. Links checked on the date shown.
- Liquid Network Security Incident Assessment — Blockstream · Official · checked
- $320M exploit of Liquid Network — Chainalysis · Analytics · checked
- Liquid Network Incident Analysis — CertiK · Security firm · checked
- DefiLlama Hacks — Liquid Network, 2026-09-06 — DefiLlama · Database · checked
How do I cite this page?
Burn, A. / upswap.io (2026). Liquid Network incident (2026-09-06). Updated 2026-10-24. https://upswap.io/incidents/liquid-network-2026#summary. CC BY 4.0. Accessed ____.
Stable anchors: #summary #facts #timeline #funds #affected #analysis #then #now #sources #cite
