CrossCurve Bridge Exploit Drains About $3M, Rekindling Cross-Chain Risk

ccn.comDipublikasikan tanggal 2026-02-02Terakhir diperbarui pada 2026-02-02

Abstrak

Cross-chain liquidity protocol CrossCurve suffered an exploit on February 2, with estimated losses around $3 million across multiple networks. The attack involved a spoofed cross-chain message that bypassed validation, allowing the attacker to trigger unauthorized token unlocks on the destination chain. The protocol urged users to pause interactions and launched an investigation. CEO Boris Povar later published ten Ethereum addresses linked to the stolen funds, offering a 10% bounty for their return within 72 hours and threatening legal action. The incident highlights persistent vulnerabilities in cross-chain bridges, where security often conflicts with user demand for speed. Verification failures and assumptions in smart contract logic remain critical risks, as a single flaw can lead to multi-network exploits.

Key Takeaways
  • CrossCurve said its bridge was “under attack” on Feb. 2 and told users to pause interactions.
  • Defimon Alerts, linked to Decurity, estimated losses around $3 million across “several networks.”
  • Early reporting and security posts described a spoofed cross-chain message that bypassed validation and triggered token unlocks on the destination chain.

Cross-chain liquidity protocol CrossCurve said its bridge was exploited on Feb. 2, with security monitors estimating roughly $3 million in losses across multiple networks.

The protocol urged users to pause interactions while it investigated.

Later, CEO Boris Povar published ten Ethereum addresses he said received funds and offered a bounty of up to 10% if the assets were returned within 72 hours, warning the project would pursue legal action if no contact was made.

Try Our Recommended Crypto Exchanges
Sponsored
Disclosure
We sometimes use affiliate links in our content, when clicking on those we might receive a commission at no extra cost to you. By using this website you agree to our terms and conditions and privacy policy.

Bitget

promotions
New user rewards up to 6,200 USDT.
Coins
88
  • Bitcoin
  • Ethereum
  • Tether
  • Litecoin
  • Bitcoin Cash
  • TRON
  • Ripple
  • Cardano
  • Binance USD
  • USD Coin
  • Polkadot
  • Shiba Inu
  • Basic Attention Token
  • Qtum
  • Ethereum Classic
  • Chainlink
  • Solana
  • Polygon Matic
  • Cosmos
  • Wrapped Bitcoin
  • PancakeSwap
  • The Sandbox
  • Storj
  • iExec RLC
  • Bancor
  • Uniswap
  • Enjin Coin
  • 1inch Network
  • Chiliz
  • Aave
  • Synthetix
  • Maker
  • Compound
  • Theta Network
  • Celo
  • Curve DAO Token
  • Decentraland
  • JUST
  • Axie Infinity
  • Gala
  • Internet Computer
  • The Graph
  • Filecoin
  • dYdX
  • Mask Network
  • Lido DAO Token
  • Reserve Rights
  • Chromia
  • Ankr
  • Ocean Protocol
  • Adventure Gold
  • Origin Protocol
  • Celer Network
  • NKN
  • Bitget Token
  • Alchemix
  • Immutable
  • Ethereum Name Service
  • Avalanche
  • Zilliqa
  • JasmyCoin
  • Toncoin
  • Convex Finance
  • Spell Token
  • Holo
  • Render Token
  • SushiSwap
  • MetisDAO
  • Audius
  • Illuvium
  • ARPA Chain
  • Osmosis
  • Fantom
  • Neo
  • Arweave
  • Algorand
  • Kusama
  • XDC Network
  • APEcoin
  • MultiversX
  • THORChain
  • NEAR Protocol
  • Nexo
  • Amp
  • Livepeer
  • PAX Gold
  • TrueUSD
  • BitTorrent
  • Golem
  • FLUX
  • Helium
  • Dai
  • Build'N'Build
No result
Claim Offer

Bitunix

promotions
Receive up to $100,000 worth of exclusive gifts for newcomers upon registration.
Coins
151
  • Bitcoin
  • Ethereum
  • Tether
  • USD Coin
  • Solana
  • Ripple
  • Dogecoin
  • Cardano
  • Toncoin
  • Shiba Inu
  • Avalanche
  • TRON
  • Chainlink
  • Polygon Matic
  • Polkadot
  • Wrapped Bitcoin
  • Litecoin
  • Dai
  • NEAR Protocol
  • Bitcoin Cash
  • Stellar
  • Cosmos
  • Filecoin
  • Ethereum Classic
  • Aptos
  • Hedera Hashgraph
  • Immutable
  • Optimism
  • Arbitrum
  • VeChain
  • The Sandbox
  • Decentraland
  • Axie Infinity
  • Injective Protocol
  • Render
  • The Graph
  • Maker
  • Aave
  • Chiliz
  • Helium
  • PAX Gold
  • Compound
  • Lido DAO Token
  • Sui
  • Conflux Network
  • Lido Staked ETH
  • OKB
  • Uniswap
  • Pepe
  • Ondo
  • Mantle
  • First Digital USD
  • XDC Network
  • Artificial Superintelligence Alliance
  • Jupiter
  • Quant
  • Worldcoin
  • Bonk
  • Tether Gold
  • JITO
  • JasmyCoin
  • Core
  • Floki Inu
  • Ethereum Name Service
  • SushiSwap
  • 1inch Network
  • Tezos
  • Algorand
  • Flow
  • Trust Wallet Token
  • Curve DAO Token
  • MultiversX
  • Basic Attention Token
  • Enjin Coin
  • Ethena
  • Ethena Staked USDe
  • Audius
  • Alchemy Pay
  • Arkham
  • API3
  • Bounce Token
  • Altlayer
  • Aergo
  • Amp
  • Aevo
  • Ankr
  • Axelar
  • Alpaca Finance
  • Blur
  • Biconomy
  • Tranchess
  • Celer Network
  • Shentu
  • Civic
  • Convex Finance
  • Cartesi
  • Cyber
  • COTI
  • DIA
  • dYdX
  • ether.fi
  • FLUX
  • Ampleforth
  • Golem
  • Holo
  • Illuvium
  • JUST
  • Livepeer
  • Memecoin
  • Manta Network
  • Treasure
  • Mask Network
  • Origin Protocol
  • ORDI
  • Ontology
  • Phala Network
  • Pendle
  • Portal
  • Pyth Network
  • ConstitutionDAO
  • iExec RLC
  • Reserve Rights
  • Ravencoin
  • Storj
  • Status
  • Spell Token
  • Sun (New)
  • SuperVerse
  • Tellor
  • LayerZero
  • Scroll
  • Usual
  • Eigenlayer
  • Hamster Kombat
  • Catizen
  • Berachain
  • KAITO
  • Pudgy Penguins
  • Solayer
  • Bio Protocol
  • ChainGPT
  • Cookie DAO
  • Solv Protocol
  • Movement
  • DeXe
  • Nexo
  • Hyperliquid
  • STEPN
  • Synthetix
  • Neo
  • APEcoin
  • Gala
  • Internet Computer
  • Pi Network
  • IoTex
  • Build'N'Build
No result
Claim Offer

BTCC

promotions
Get up to 10,055 USDT when you register, verify, and make the first deposit and the first trades.
Coins
162
  • Ethereum
  • Tether
  • USD Coin
  • Solana
  • Ripple
  • Dogecoin
  • Cardano
  • Toncoin
  • Shiba Inu
  • Avalanche
  • TRON
  • Chainlink
  • Polygon Matic
  • Polkadot
  • Litecoin
  • NEAR Protocol
  • Bitcoin Cash
  • Stellar
  • Cosmos
  • Filecoin
  • Ethereum Classic
  • Aptos
  • Immutable
  • Optimism
  • Arbitrum
  • VeChain
  • The Sandbox
  • Decentraland
  • Axie Infinity
  • Injective Protocol
  • The Graph
  • Hedera Hashgraph
  • Render Token
  • Aave
  • Chiliz
  • PAX Gold
  • Compound
  • Lido DAO Token
  • THORChain
  • Stacks
  • Arweave
  • Sui
  • Conflux Network
  • Uniswap
  • Pepe
  • Ondo
  • Mantle
  • First Digital USD
  • Bittensor
  • Kaspa
  • Celestia
  • Artificial Superintelligence Alliance
  • Jupiter
  • Quant
  • Worldcoin
  • PayPal USD
  • Bonk
  • Rocket Pool ETH
  • Tether Gold
  • Sei
  • JITO
  • JasmyCoin
  • PancakeSwap
  • Floki Inu
  • Ethereum Name Service
  • SushiSwap
  • 1inch Network
  • Algorand
  • Flow
  • Trust Wallet Token
  • Curve DAO Token
  • Basic Attention Token
  • Enjin Coin
  • Ethena
  • Ethena USDe
  • Pi Network
  • Adventure Gold
  • Audius
  • Acala Token
  • Alchemy Pay
  • Arkham
  • API3
  • Bounce Token
  • Altlayer
  • Amp
  • Aevo
  • ARPA Chain
  • Ankr
  • Blur
  • Biconomy
  • Chromia
  • Celer Network
  • Celo
  • Civic
  • Convex Finance
  • Cartesi
  • COTI
  • DigiByte
  • DIA
  • Dymension
  • dYdX
  • ether.fi
  • FUNToken
  • FLUX
  • Ampleforth
  • Golem
  • GMX
  • Holo
  • IoTex
  • Illuvium
  • JUST
  • Liquity
  • Livepeer
  • Memecoin
  • Manta Network
  • Treasure
  • Mask Network
  • NKN
  • Neutron
  • Ocean Protocol
  • Origin Protocol
  • ORDI
  • Osmosis
  • Powerledger
  • Phala Network
  • Pendle
  • Portal
  • Pyth Network
  • ConstitutionDAO
  • iExec RLC
  • Rocket Pool
  • Reserve Rights
  • Storj
  • Starknet
  • Spell Token
  • Sun (New)
  • Saga
  • SuperVerse
  • Toko Token
  • Tellor
  • LayerZero
  • Usual
  • Cetus Protocol
  • Eigenlayer
  • Hamster Kombat
  • Neiro - First Neiro on Ethereum
  • Catizen
  • Berachain
  • KAITO
  • Pudgy Penguins
  • Solayer
  • Alchemix
  • Bitcoin
  • Bitcoin SV
  • Movement
  • Nexo
  • Hyperliquid
  • Nervos Network
  • TrueUSD
  • Mina
  • STEPN
  • Synthetix
  • APEcoin
  • Gala
  • Cronos
  • Internet Computer
  • Build'N'Build
No result
Claim Offer
Explore All Offers

CrossCurve Attack Timeline

CrossCurve said on Feb. 2 that its bridge was “under attack,” involving exploitation of a vulnerability in one of the smart contracts used in its cross-chain system.

The exploit allowed an attacker to spoof a message to bypass validation and unlock tokens.

One quoted description said an attacker could call an “express” execution path on a receiver contract using a forged cross-chain message, then trigger an unlock on a portal contract.

CrossCurve has not published a full post-mortem or confirmed a final loss figure. Separate estimates clustered around $3 million.

In a follow-up post, Povar said the team identified ten Ethereum addresses tied to received funds and set a 72-hour window to return assets or make contact before escalation.

He said the project was prepared to pursue civil and criminal remedies and coordinate with industry partners to freeze assets.

CrossCurve did not immediately respond to a request for comment on the specific bug, the final loss amount, or a timeline for reopening.

A separate warning came from Curve Finance, which said users allocated to CrossCurve pools “may wish to review their positions” and consider removing votes, urging “risk-aware decisions” when interacting with third parties.

Why Spoofed Messages and Validation Assumptions Keep Winning

Bridge exploits often look like “just a smart contract bug.” The deeper pattern is verification failure.

A bridge is a promise: release assets on Chain B because something real happened on Chain A. The hard part is proving that “something real” without trusting an attacker’s message.

In general message passing, the destination contract is supposed to verify that a call was approved by the validator set by checking with the gateway (for example, via a validation function) before executing.

If a receiver contract accepts an alternate path that skips or weakens that check, a forged message can become a payout.

That’s why the “receiver side” matters as much as the messaging layer.

A protocol can route messages through reputable infrastructure and still lose funds if its own destination contract implements permissive logic, unsafe fast paths, or incorrect assumptions about upstream guarantees.

CrossCurve’s own documentation frames cross-chain risk as a “black swan” category and describes a design goal of routing through multiple independent validation protocols (“Consensus Bridge”) to reduce single points of failure.

But even multi-path designs can be undermined by a weak integration contract at the edge.

The Uncomfortable Truth: Bridge UX Wants Speed, Security Wants Paranoia

Users want bridging to feel instant: fewer clicks, less waiting, faster finality.

Security wants the opposite: more confirmations, tighter limits, and “do nothing unless you’re sure.”

Some cross-chain stacks explicitly offer speed features like “express” execution, where off-chain actors can accelerate delivery of an intended outcome.

The trade-off is that fast paths demand extra care in how authenticity is enforced, because the system is trying to move before the slowest proofs arrive.

This tension is why bridge hacks stay evergreen. Bridges concentrate liquidity, and a single verification bypass can unlock assets across multiple networks in one run.

What To Watch Next

CrossCurve has not yet released a full incident report. In most bridge incidents, the next signals that matter are:

  • Whether contracts remain paused and what code changes ship before any restart.
  • Whether the attacker returns funds, often in exchange for a bounty.
  • Whether stablecoin issuers, exchanges, or analytics firms flag and freeze related addresses.
  • Whether independent security teams publish a corroborated root-cause analysis.

For now, the takeaway is familiar and still useful: cross-chain bridges remain one of crypto’s most repeatable failure points, because “truth across chains” is a hard engineering problem with real money behind every assumption.

This is a developing story and will be updated.

Recommended Secure Partners
  • Safest Exchanges Best Safest (Most Secure) Crypto Exchanges? Check Out These Exchanges
  • Secure Crypto Wallets Crypto Wallets Reviews and Ranked
  • Bet Anonymously Check Out Our Recommended No KYC Casinos

Pertanyaan Terkait

QWhat was the estimated financial loss from the CrossCurve bridge exploit?

AThe estimated financial loss from the CrossCurve bridge exploit was approximately $3 million across several networks.

QWhat was the technical cause of the CrossCurve exploit as described in early reports?

AThe exploit was caused by a spoofed cross-chain message that bypassed validation, which then triggered unauthorized token unlocks on the destination chain.

QWhat action did CrossCurve's CEO take in response to the attack?

ACrossCurve's CEO, Boris Povar, published ten Ethereum addresses that received the funds and offered a bounty of up to 10% if the assets were returned within 72 hours, warning of legal action if no contact was made.

QAccording to the article, what is the fundamental tension that makes bridge exploits a recurring problem?

AThe fundamental tension is that users want bridging to be fast and instant, while security requires more confirmations, tighter limits, and cautious verification, creating a conflict between user experience and security paranoia.

QWhat general warning did Curve Finance issue in relation to this incident?

ACurve Finance warned users allocated to CrossCurve pools to review their positions and consider removing votes, urging them to make 'risk-aware decisions' when interacting with third parties.

Bacaan Terkait

Laporan Pagi Wall Street: V-reversal di Akhir Bulan, Tapi Indeks Nasdaq Rekam Performa Terburuk Juli dalam 12 Tahun, Modal Terkonsentrasi ke Raksasa Cloud

**Laporan Pagi Wall Street: Rebound di Akhir Bulan, Tapi Nasdaq Catat Performa Terburuk Juli dalam 12 Tahun, Modal Terakselerasi ke Raksasa Cloud** Pasar saham AS mengalami rebound berbentuk V pada Jumat lalu, dengan tiga indeks utama menguat. Namun, ini tidak menutupi performa buruk sepanjang Juli: Indeks S&P 500 hampir datar, mencatat kinerja Juli terburuk sejak 2014, sementara **Indeks Nasdaq turun 3,2%, merupakan kinerja Juli terburuk sejak 2004.** **Pergerakan Komoditas & Makro:** * **Minyak:** Harga minyak mentah WTI anjlok >8% setelah Presiden Trump membatalkan serangan militer terhadap Iran dan OPEC+ menyetujui peningkatan kuota produksi. * **Emas:** Harga emas naik tipis 0,91% pada Juli, berayun di sekitar $4050/oz. Analis melihat periode konsolidasi dengan kisaran perkiraan $3800-$4500 untuk tahun ini. * **Obligasi & Mata Uang:** Yield obligasi AS 10 tahun melonjak di bulan Juli. AS dan Jepang melakukan intervensi bersama untuk mendukung Yen, mendorong USD/JPY turun di bawah level 156. **Fokus Pasar: Percepatan Aliran Modal ke Raksasa Cloud** Tekanan deleveraging membayangi saham teknologi sepanjang Juli, namun **modal bergerak cepat ke perusahaan cloud besar.** Saham cloud meroket setelah laporan kinerja kuartalan yang kuat, dengan **kapitalisasi pasar gabungan raksasa cloud bertambah hampir $1,5 triliun** dalam seminggu (Microsoft, Amazon, Google). Ini menunjukkan pergeseran modal dari sektor hardware/semikonduktor yang padat menuju perusahaan cloud dengan arus kas yang kuat. **Kilasan Performa Saham Penting:** * **Amazon:** Melonjak 15,32% setelah kinerja AWS yang sangat kuat. * **Microsoft & Google:** Terus menguat dengan prospek bisnis cloud/AI yang solid. * **Meta:** Naik 3,28%, mengakhiri 11 hari penurunan beruntun. * **Apple:** Anjlok 7,35% karena kekhawatiran pasokan dan panduan yang lemah. * **Saham Memori (seperti Micron, Sandisk):** Terkoreksi tajam pada bulan Juli. * **SpaceX:** Akan merilis laporan kinerja pertama pasca-IPO pada 5 Agustus, diikuti masa pelepasan saham terbatas besar-besaran pada 6 Agustus. **Yang Perlu Diperhatikan Minggu Ini:** * **Laporan Kinerja Perusahaan:** Berkshire Hathaway, Palantir, AMD, SpaceX, Disney, Sandisk, Western Digital, dan lainnya. * **Konferensi:** Ai4 2026 dan Flash Memory Summit (FMS). * **Data Makro Kunci:** Laporan Penggajian Non-Pertanian AS (7 Agustus), data ekspor-impor & CPI China.

marsbit45m yang lalu

Laporan Pagi Wall Street: V-reversal di Akhir Bulan, Tapi Indeks Nasdaq Rekam Performa Terburuk Juli dalam 12 Tahun, Modal Terkonsentrasi ke Raksasa Cloud

marsbit45m yang lalu

Rubin Ultra Terkuras Drastis, Apakah Nvidia Juga Tak Tahan dengan Kenaikan Harga Memori?

Laporan terbaru dari SemiAnalysis mengungkapkan bahwa NVIDIA telah merevisi spesifikasi Rubin Ultra, varian flagship dari arsitektur Rubin yang diumumkan di GTC 2026. Alih-alih peningkatan besar pada memori dan daya komputasi, fokus utama kini beralih ke kemampuan interkoneksi skala sistem. Spesifikasi kunci yang diungkapkan menunjukkan Rubin Ultra akan mempertahankan puncak komputasi teoretis yang sama dengan Rubin standar, yaitu 35 PFLOPs. Namun, terjadi pengurangan signifikan pada kapasitas memori, turun menjadi 192GB dengan konfigurasi 8-Hi, lebih rendah dari Rubin standar yang 288GB (12-Hi). Peningkatan bandwidth memori juga minimal, hanya 1 TB/s, dan daya chip bahkan sedikit lebih tinggi. Peningkatan utama justru pada "skala interkoneksi" (*scale-up world size*), yang melonjak dari 72 GPU menjadi 576 GPU. Ini memungkinkan ratusan GPU Rubin Ultra terhubung via NVLink membentuk satu "super GPU" logis yang sangat besar. Analisis SemiAnalysis mengaitkan perubahan desain ini dengan lonjakan harga HBM (High Bandwidth Memory) yang drastis. Biaya material (BOM) untuk rak server Rubin Ultra sempat melonjak menjadi sekitar $8 juta karena harga HBM, tetapi dengan spesifikasi baru, biaya turun ke sekitar $6.4 juta. NVIDIA diduga mengalihkan biaya dari komponen HBM yang mahal (yang pangsa biayanya turun dari ~40% menjadi 28%) ke peningkatan interkoneksi jaringan (naik dari 4% menjadi 12%). Berita ini memicu kekhawatiran pasar tentang permintaan HBM. Saham produsen HBM seperti SK Hynix dan Samsung anjlok sekitar 8%, mencerminkan kekhawatiran bahwa pembeli utama seperti NVIDIA mungkin mulai mengurangi ketergantungan pada HBM berkapasitas tinggi untuk mengontrol biaya sistem AI secara keseluruhan.

Odaily星球日报1j yang lalu

Rubin Ultra Terkuras Drastis, Apakah Nvidia Juga Tak Tahan dengan Kenaikan Harga Memori?

Odaily星球日报1j yang lalu

Model Generatif Bisa Dilatih End-to-End? Intinya Hanya Sebuah For Loop

Model generatif seperti model difusi atau autoregresif yang dominan saat ini biasanya tidak melatih seluruh proses generasi secara end-to-end. Mereka dilatih hanya untuk memprediksi satu "langkah kecil" dalam proses, namun saat inferensi, langkah ini harus diulang ratusan atau ribuan kali. Ketidaksesuaian ini menyebabkan masalah "exposure bias," di mana kesalahan menumpuk seiring waktu. Makalah terbaru dari UIUC dan Harvard memperkenalkan paradigma baru bernama **Explorative Modeling (XM)**. Inti dari metode ini sangat sederhana: alih-alih menghasilkan satu sampel, model menghasilkan **K kandidat** dalam setiap langkah pelatihan. Kemudian, hanya kandidat yang paling dekat dengan data target yang dipilih untuk menghitung gradien dan memperbarui model. Mengapa ini berhasil? Dalam tugas generatif, satu input bisa memiliki banyak output yang valid (mode). Fungsi kerugian rekonstruksi tradisional cenderung menyebabkan "mode blurring," di mana model hanya mempelajari rata-rata dari semua mode, menghasilkan output yang tidak realistis. Dengan menghasilkan beberapa kandidat, model dapat "mengeksplorasi" dan menangkap beberapa mode yang berbeda sekaligus, sehingga menghindari rata-rata yang kabur. Penelitian ini menunjukkan bahwa **"eksplorasi" (K) bertindak sebagai sumbu penskalaan ketiga** yang efektif, selain parameter dan data. Eksperimen pada gambar, video, dan bahasa menunjukkan peningkatan kinerja yang monoton dengan peningkatan K, dengan keuntungan yang lebih besar pada skala yang lebih besar. XM dapat meningkatkan efisiensi FLOP hingga 4.1x dan efisiensi sampel hingga 6.2x. Yang lebih menarik, ketika diterapkan secara ekstrem, XM memungkinkan **pelatihan generatif yang benar-benar end-to-end**. Dalam tugas kontrol robot, "Explorative Policy" berbasis XM mencapai kinerja yang setara dengan "Diffusion Policy" yang membutuhkan 100 langkah inferensi, sementara hanya membutuhkan satu langkah inferensi maju. Ini membuktikan bahwa kompleksitas untuk menangani banyak mode dapat dipindahkan dari inferensi ke pelatihan. Meskipun ide "best-of-K" bukan hal baru, kontribusi utama makalah ini adalah penjelasan teoritis bahwa mekanisme ini secara langsung meningkatkan **ekspresivitas generatif** model. Dengan demikian, ini menawarkan cara baru yang ampuh untuk mengatasi tantangan mendasar dalam pemodelan generatif.

marsbit2j yang lalu

Model Generatif Bisa Dilatih End-to-End? Intinya Hanya Sebuah For Loop

marsbit2j yang lalu

Trading

Spot
活动图片