ThreatsDay: AI-Powered Zero-Day Chain, 543K Live Secrets, Model Inspection RCE and 13 More Stories
Weekly ThreatsDay digest: OFAC sanctions Tren de Aragua ATM jackpotting network, blockchain malware dead drops surge, plus cache-poisoning and EDR-evasion research.
The roundup covers OFAC sanctions on 10 Tren de Aragua targets behind $40.73 million in Ploutus-based ATM jackpotting losses across more than 1,500 U.S. attacks, and Chainalysis reporting a 440% surge in blockchain dead drop (EtherHiding) malware delivery. It also summarizes Moonshot's internal AI safety review after Mindgard found Kimi K2.6 and K3 Swarm could bypass guardrails, Tracebit's Context Bombs prompt-injection defense, and YesWeHack's cache key injection poisoning technique. Additional items include a named-pipe process injection EDR evasion technique and a Huntress incident where attackers compiled a cryptominer on the victim host after exploiting Samsung MagicINFO (CVE-2025-4632).
- OFAC sanctioned 10 Tren de Aragua targets over $40.73M ATM jackpotting losses
- Blockchain dead drop malware delivery surged 440%, per Chainalysis
- Moonshot reviewing Kimi K2.6/K3 Swarm guardrail bypass findings
- New research: console named-pipe EDR evasion, cache key injection
- Huntress documents on-host cryptominer compile after MagicINFO exploit
- CVEs
- CVE-2025-4632
- Tags
- threat-intelweekly-roundupsanctionsatm-malwareploutusblockchain-dead-dropprompt-injectioncache-poisoning
Vulnerabilities mentioned
All| CVE | Vulnerability | CVSS | EPSS | Signals | Published |
|---|---|---|---|---|---|
| CVE-2025-4632 | Actively Exploited Path Traversal File Write in Samsung MagicINFO 9 Server | 9.8 | 22% | KEV | KEV |
From the original article
10 min read at thehackernews.com
This week, the useful words are boring ones: inspect, cache, compile, store, trust. Each sounds harmless. Each can become an attack path when a system does a little more than people expect.
thehackernews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.