Threat Brief — 2026-08-22 — Banking trojan widens net

A banking trojan now targets 349 financial and crypto-wallet apps via overlay attacks and touch-event monitoring, representing a broadening of the mobile threat surface for end users. Separately, a crowdsourced ML project demonstrated that Gemini can recognise water levels from citizen photos with 98% accuracy — notable for data-handling implications but not a direct security threat.

Top items

• Banking trojan targeting 349 apps via overlay and touch-monitoring attacks. A new version of a mobile banking trojan monitors screen touches, substitutes app windows with fake overlays, and can attack hundreds of banking and cryptocurrency wallet applications. The breadth of targeted apps — 349 — significantly raises the risk of credential theft and fraudulent transactions for users of affected financial institutions. This story was first reported today. (src: RSS:securitylab-ru)

• Gemini ML model recognises water levels from crowdsourced photos at 98% accuracy. Over 15 years, 8,000 volunteers photographed streams; Gemini now parses those images through a black-and-white filter to determine water levels without retraining from scratch. While not a security incident, the reliance on large-scale crowdsourced image data and cloud-based AI inference raises data-privacy and model-integrity considerations for organisations building similar pipelines. (src: RSS:securitylab-ru)

Themes

Mobile banking malware consolidation. The 349-app banking trojan continues a recent pattern of Android malware expanding scope — from credential theft to full overlay injection and touch-event capture — underscoring that MDM and app-pinning controls remain underused in many environments.

===