Worked example · 2021-12

AscendEX hot-wallet compromise — multi-chain (Ethereum, BSC, Polygon) — 2021-12-11

Loss: approximately $77.7M extracted from AscendEX hot wallets across Ethereum (~$60M), Binance Smart Chain (~$9.2M), and Polygon (~$8.5M) in a single attack window on 2021-12-11. Asset breakdown per Peckshield and SlowMist contemporaneous forensic reconstruction: ETH, USDT, USDC, SHIB, MATIC, FTM, and other ERC-20 / BEP-20 / Polygon-native tokens.
Recovery: partial — stablecoin issuers (Tether, Circle) froze a portion of the USDT/USDC proceeds on the day of the breach; the bulk of the non-stablecoin proceeds was laundered through DEX swaps and mixer rails before freezes propagated. AscendEX absorbed the residual operational shortfall and resumed withdrawals within approximately three days.
OAK Techniques observed: OAK-T4.003 (Private-Key Theft — the attacker gained access to AscendEX's hot-wallet private-key material across Ethereum, BSC, and Polygon simultaneously; the multi-chain simultaneous extraction pattern indicates co-located key material in the operator-side architecture). OAK-T11.001 (Third-Party Signing / Custody Vendor Compromise — the attacker compromised AscendEX's hot-wallet signing infrastructure, gaining access to private-key material across Ethereum, BSC, and Polygon hot wallets simultaneously; the multi-chain simultaneous extraction pattern indicates co-located key material in the operator-side architecture). OAK-T15.003 (Operator Endpoint / Infrastructure Compromise — the off-chain entry vector was intrusion into the exchange's hot-wallet infrastructure; the precise intrusion mechanism was not publicly disclosed by AscendEX, but the broader DPRK / TraderTraitor 2021 cohort consistently exhibits the social-engineering-of-operator-personnel plus endpoint-compromise pattern documented by Chainalysis and Elliptic). Downstream OAK-T7.001 (Mixer-Routed Hop — Tornado Cash for the ETH / ERC-20 proceeds) and OAK-T7.003 (Cross-Chain Bridge Laundering — the attacker rotated non-ETH-chain assets through cross-chain bridges and DEX swaps before mixer routing). OAK-T8.001 (Cluster-Reuse — the wallet-cluster analysis underpinning the Lazarus / DPRK attribution is the load-bearing operator-attribution primitive; the AscendEX cluster overlaps with the confirmed-DPRK-attributed wallet clusters documented in the April 2022 FBI / CISA / Treasury joint advisory).
Attribution: inferred-strong Chainalysis and Elliptic published wallet-cluster attribution to DPRK-aligned operators (the Lazarus / TraderTraitor cluster) as part of the broader 2021 DPRK-theft aggregate (~$400M across 7+ events). The April 2022 FBI / CISA / Treasury joint advisory on TraderTraitor consolidates this cluster-level attribution but does not single out AscendEX by name. By OAK convention the case is inferred-strong, not confirmed, because no FBI / DOJ / Treasury public statement specifically names AscendEX as a Lazarus-attributed event at v0.1 cutoff. OAK records the G01 cluster attribution in narrative form.
OAK-Gnn: OAK-G01 Lazarus Group / DPRK-attributed — inferred-strong, cohort-level.
Key teaching point: Multi-chain hot-wallet key-store co-location produces multi-chain loss surfaces from a single compromise event. AscendEX's December 2021 incident drained simultaneously across Ethereum, BSC, and Polygon, indicating shared or co-located signing infrastructure. The pattern recurs across the broader DPRK TraderTraitor cluster (KuCoin 2020, Liquid 2021, Poloniex 2023, Phemex 2025) and is canonical for OAK-T11 contributions discussing multi-chain key-isolation as a discrete custody-side architectural variable.

Summary

AscendEX (formerly BitMax) is a centralised exchange headquartered in Singapore. On December 11, 2021, an attacker compromised the exchange's hot-wallet infrastructure and drained approximately $77.7M in cumulative value across Ethereum, BSC, and Polygon hot wallets in a single window. The simultaneous multi-chain extraction pattern indicates that the underlying signing-infrastructure compromise was at a single point in the operator-side architecture rather than per-chain isolated compromises.

Per Peckshield and SlowMist on-chain analysis, the stolen assets included approximately $60M in ETH and ERC-20 tokens on Ethereum, $9.2M in BEP-20 tokens on BSC, and $8.5M in tokens on Polygon. The attacker rapidly swapped non-stablecoin assets through DEXs and routed the proceeds through Tornado Cash and cross-chain bridges. Tether and Circle froze a portion of the stablecoin proceeds within hours; the residual non-stablecoin balance was not recoverable via issuer-side freeze.

AscendEX paused withdrawals, migrated remaining wallet balances to cold storage, and resumed operations within approximately three days, absorbing the operational shortfall. AscendEX did not publicly disclose the precise intrusion mechanism. No DOJ / civil-forfeiture action specifically naming AscendEX has been filed at v0.1 cutoff.

The case structurally pairs with Liquid Global August 2021 (/examples/2021-08-liquid-global.md) as the two canonical 2021 G01-attributed exchange hot-wallet compromises, and is a precursor to the larger 2023-2025 exchange-compromise cluster (Poloniex → DMM → WazirX → Phemex → Bybit) that represents the primary OAK-G01 TraderTraitor surface.

Timeline (UTC)

When	Event	OAK ref
2021-12-11	Attacker compromises AscendEX hot-wallet signing infrastructure; simultaneous multi-chain extraction across ETH / BSC / Polygon; ~$77.7M cumulative outflow	T11.001 + T15.003 (extraction + entry-vector)
2021-12-11 (within hours)	AscendEX pauses withdrawals; some stablecoin balances frozen by Tether / Circle	(operator response)
2021-12-11 to 2021-12-12	Attacker swaps non-stablecoin assets through DEXs; routes proceeds through Tornado Cash and cross-chain bridges	T7.001 + T7.003
2021-12-12 to 2021-12-14	Peckshield, SlowMist publish on-chain trace and cumulative-loss aggregation	(forensic record)
2021-12-14	AscendEX resumes withdrawals; operational shortfall absorbed by operator	(operator recovery)
2022-04	FBI / CISA / Treasury joint TraderTraitor advisory consolidates cluster-level DPRK attribution for the 2021 cohort; does not name AscendEX specifically	G01 inferred-strong attribution
Continuing	No DOJ / civil-forfeiture action specifically naming AscendEX	(attribution state)

What defenders observed

Simultaneous multi-chain extraction as single-point-of-compromise diagnostic. The simultaneous drain across Ethereum, BSC, and Polygon indicates co-located hot-wallet key material. The pattern is identical to KuCoin 2020 and Liquid Global 2021 and recurs through the 2023-2025 TraderTraitor cluster. Defender architectural review should treat this pattern as a high-confidence indicator of single-point compromise in the operator-side custody architecture.
Multi-chain key-store co-location is a discrete custody-side architectural anti-pattern. The mitigation surface is per-chain key isolation via HSM/MPC architecture (OAK-M37) combined with air-gap signing for the cold-wallet portion of the treasury (OAK-M19).
Stablecoin-issuer freeze is a partial but time-gated recovery lever. Tether and Circle froze a portion of the proceeds within hours; the non-stablecoin balance had already been swapped through DEXs and was unrecoverable via issuer-side action. The freeze window narrows as the attacker's DEX-swap cadence increases.

What this example tells contributors writing future Technique pages

T11.001 + T15.003 is the canonical exchange-compromise technique chain. Future T11 contributions documenting exchange hot-wallet compromises should expect the T15.003 (Operator Endpoint Compromise) entry-vector pairing and should record whether the off-chain intrusion mechanism was publicly disclosed.
Multi-chain extraction is a structural signal, not an incident-type classifier. The same multi-chain extraction pattern appears across G01-attributed (KuCoin, Liquid, AscendEX, Poloniex) and non-G01-attributed (BitMart) exchange compromises. Contributors should treat the pattern as an architectural diagnostic rather than as a per-incident-specific observation.

Public references

[peckshieldascendex2021] — Peckshield on-chain trace and cumulative-loss aggregation for the 2021-12-11 AscendEX hot-wallet compromise.
[slowmistascendex2021] — SlowMist incident analysis and laundering-route tracing.
[coindeskascendex2021] — CoinDesk coverage of the breach and operator response.
[chainalysisdprk2022] — Chainalysis 2022 DPRK report; includes AscendEX in the 2021 DPRK-theft aggregate.
[ellipticascendex2021] — Elliptic wallet-cluster attribution to DPRK-aligned operators.
[fbicisatreasurytrader2022] — FBI / CISA / Treasury April 2022 joint TraderTraitor advisory; cluster-level attribution.

Discussion

AscendEX December 2021 is structurally the companion case to Liquid Global August 2021 as the two canonical 2021 G01-attributed exchange hot-wallet compromises. The two cases share the same technique chain (T15.003 → T11.001 → T7.001 + T7.003 → T8.001), the same attribution cluster (inferred-strong DPRK / TraderTraitor), and the same architectural observation (multi-chain key-store co-location as a custody-side anti-pattern). The pair closes the T11 2021 and T15 2021 near-threshold coverage gaps at v0.1.

The case is included at the "valid but not exhaustively elaborated" level of detail consistent with the v0.1 standard for long-tail exchange-compromise incidents. Future v0.x updates may expand the forensic detail as additional public-record material (regulatory filings, industry-forensic retrospectives) becomes available.