2024 · Indonesia · K - Financial and insurance activities · Cryptocurrency theft from hot wallets via compromised withdrawal path
Analysed 2026-11-06 by Hermes
Open this case in the tool →Cryptocurrency theft from hot wallets via compromised withdrawal path
Expertise to Execution
Evidence coverage: 3/6 stages, weight 20/60
Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution
(1) As Indonesia's largest licensed crypto exchange (a critical-infrastructure finance target per the EuRepoC record), the expectation at Indodax was that its crown-jewel custody controls - access control over the withdrawal-authorisation path, anomaly monitoring of hot-wallet outflows and egress interruption - actually worked in production. (2) In practice, from at least 10 September 2024 an attacker issued 150+ seemingly legitimate withdrawals across five chains (SlowMist observed whole-amount withdrawals with change addresses routed back into Indodax infrastructure to mask the activity) and drained about USD 20.5-22 million from the hot wallets before PeckShield, Cyvers and SlowMist raised the first public alerts on 11 September; Cyvers's real-time system flagged 160 critical red flags at the onset and its head of AI described the vector as related to an access control violation, while the CEO acknowledged security lapses and closed the exploited loopholes only after the theft. (3) The classification rests on the entry stage, which carries the clearest origin/break documentation: the failed expectation is specialist technical control knowledge of a crypto exchange (material origin) and the exploited hole sat unremediated in the live platform until after the breach (execution break) - expertise available but not realised, i.e. Expertise to Execution. The exact entry mechanism is disputed (withdrawal system per SlowMist, signature machine per Cyvers) and is not load-bearing; what matters is that a control hole in the victim's own withdrawal path enabled the theft. The detection gap (external watchdogs noticed first after about a day of dwell) and the egress gap (value left without internal interruption) point to the same type and corroborate it; no source documents a pre-breach policy, standard or resourced programme whose non-execution caused the failure, which rules out the Strategy columns as the decisive reading (Strategy to Execution remains the runner-up). Recovery and response show aligned conduct (assets moved to cold storage, users reimbursed, regulators and police engaged). RM29 (Expertise to Execution lens): the platform's own withdrawal-authorisation path failed on access control - Cyvers described the vector as related to an access control violation and the victim acknowledged exploitable loopholes that were remediated only after the theft - so secure-design practice for the exchange's crown-jewel process was not realised in its code and architecture (causal role: Enabling). RM34 (Expertise to Execution lens): internal monitoring was insufficiently configured, staffed or tuned to raise the alarm, since 150+ suspicious transfers across five chains over about a day were surfaced by Cyvers, PeckShield and SlowMist rather than by Indodax's own alerting (causal role: Obscuring). RM35 (Expertise to Execution lens): roughly USD 20.5 million exited the hot wallets across chains within hours to a day without any internal egress interruption, while Cyvers's monitoring showed the same flow pattern was detectable and interruptible in real time (causal role: Amplifying).
RM29 — Secure-development / non-production exposureRM34 — Monitoring / detection / alert-handling gapRM35 — Egress / exfiltration monitoring gapMechanism unconfirmed: SlowMist pointed to a breach of the withdrawal system, Cyvers suggested other systems such as the signature machine were also attacked, and none of the sources reviewed discloses a publicly established root cause; Indodax itself never published a loss figure (the CEO referred only to potential losses of up to several hundred billion rupiah, so the USD 20-22 million figures are vendor estimates). Detection account is mixed: the CEO said Indodax's cybersecurity team found signs of unauthorized access and security lapses, but the first public alerts came from external security firms on 11 September and the platform's shutdown was initially framed as complete maintenance. Attribution to a DPRK-affiliated actor rests on an analysis by a crypto security agency assisting Indodax (CEO statement via ANTARA and Kompas) plus Cyvers's pattern comparison to Lazarus Group; no government has formally attributed the theft. The X.com posts listed in the case file's source_url could not be opened (login wall); those statements were read as quoted in the cited news articles. EuRepoC fields are internally inconsistent with the narrative: intelligence_impact says no data breach/exfiltration while the hijacking/economic fields and the description record the USD 20 million theft; the description does match the stated victim. Link check: cointelegraph.com was unreachable at check time (timeout); the page was opened and read during research.