--- title: "Ethereum patches Glamsterdam deposit flaw that could have diverted staking ETH" description: "Ethereum developers merged a fix for a flaw in the upcoming Glamsterdam upgrade that could have let an operator withdraw ETH staking services deposited for…" language: en canonical: "https://whycoinmoves.com/en/events/sentient-ethereum-fixes-glamsterdam-bug-that-24562d6034ca68" published: 2026-10-27T12:28:58.000Z updated: 2026-10-27T13:31:20.000Z indexable: false --- # Ethereum patches Glamsterdam deposit flaw that could have diverted staking ETH Ethereum developers merged a fix for a flaw in the upcoming Glamsterdam upgrade that could have let an operator withdraw ETH staking services deposited for… Attributed sources: [CoinDesk]() ## Key facts Ethereum developers merged a fix for a flaw in the upcoming Glamsterdam upgrade that could have let an operator withdraw ETH staking services deposited for new validators, CoinDesk reported. Researcher Avsetsin described the flaw, which depended on how Lido, Origin and Rocket Pool handle deposits; no losses were reported. The fix was first rejected on October 19 and merged Monday. Evidence strength: Single source, 1 publisher group(s) (CoinDesk). After site detection, observed returns were 5 minutes: +0.27% (difference from BTC +0.24 pp; Oct 27, 2026, 12:28 → Oct 27, 2026, 12:33, UTC; Gate · USDT); 1 hour: +2.54% (difference from BTC +2.67 pp; Oct 27, 2026, 12:28 → Oct 27, 2026, 13:31, UTC; Gate · USDT) In the preceding 30 days, the 99th percentile of absolute hourly returns was 3.35% (OKX · USDT; cutoff Oct 27, 2026, 12:28, UTC). ## Returns after site detection | Window | Return | BTC | Difference from BTC | Actual sampled interval (UTC) | Source | | --- | --- | --- | --- | --- | --- | | 5 minutes | +0.27% | +0.02% | +0.24 pp | Oct 27, 2026, 12:28 → Oct 27, 2026, 12:33 | Gate · USDT | | 1 hour | +2.54% | -0.13% | +2.67 pp | Oct 27, 2026, 12:28 → Oct 27, 2026, 13:31 | Gate · USDT | ## Historical hourly baseline | Metric | Value | | --- | --- | | Coverage | 99.3% (715/720) | | Median absolute hourly return | 0.73% | | 99th percentile of absolute hourly returns | 3.35% | | BTC beta | 1.12 | | Source / cutoff (UTC) | OKX · USDT / Oct 27, 2026, 12:28 | ## Claims and open questions - Researcher Avsetsin wrote that under draft rules an operator who registers as a builder with a particular public key could have both a builder deposit and a staking service's validator deposit credited to the builder, allowing the operator to withdraw the protocol's funds. - Avsetsin called the findings "conditional code findings, not confirmed live exploits" and said existing checks rejecting an earlier builder deposit could stop the attack; no losses have been reported. - The fix was closed on October 19 because there was not enough client support, then reopened after further review and merged by Justin Traglia, who maintains Ethereum's consensus specifications. - Will the postponed October 26 Glamsterdam launch on the Hoodi test network get a new date, which Ethereum needs before a mainnet launch can be scheduled? - When will client teams build the consensus-specification change into their software, and would live deployments handle same-key builder and validator deposits as the patched rules intend? ## Source evidence - [Ethereum fixes Glamsterdam bug that could have sent staking deposits to builders]() — CoinDesk; published Oct 27, 2026, 07:09; received Oct 27, 2026, 07:11 (UTC). ## Recent SENT events and market reactions No other publication-ready SENT events yet. - [Full coin archive]() ## Recent General news events on other coins - [PI: Pi Network releases Pi Desktop 0.6.4 with simpler SoloHost remote access]() - [ZEC: Zcash Foundation appoints David Campbell as executive director]() - [BTC: Bitcoin trades below $80,000 as analysts flag near-term weakness]() - [General news]() Evidence levels describe source coverage, not the probability that a report is true. Price moves and their timing cannot establish a cause; minute samples show an observed interval and alert thresholds are heuristics. Explanations are model-assisted, with citations and figures checked. Source titles stay in their original language; full articles are not republished. [Methods and limitations]()