Mark RadarMARK RADAR
About
EN
Sign in

微軟揭露「ASCII 走私」技術遭轉用於百萬級釣魚攻擊

1 reports · First detected 2026-09-07 · Last active 2026-09-07

微軟揭露源自 AI 安全領域的「ASCII 走私」技術已被駭客廣泛移植至傳統網路釣魚郵件中。攻擊者利用不可見的 Unicode 標籤字元拆解金融誘餌關鍵字以繞過系統過濾,單日最高發送達 237 萬封。此事件顯示 AI 時代的新興防護規避手段已對整體資安防線構成實質威脅。

All Coverage

1 original reports

The Backstory

The history behind this event
Microsoft Warns Hackers Are Exploiting OAuth Redirects to Intercept Credentials in Man-in-the-Middle Attacksfirst seen 2026-03-04 · 1 reports · similarity 0.68 · same topic: Microsoft

OAuth is an authorization standard widely used by websites and cloud services. After users sign in, they are redirected to a preset URL, allowing third-party applications to obtain limited access. Microsoft’s security team said attackers are exploiting gaps in the process’s validation controls to lure victims to spoofed pages, posing a particular concern for governments and public-sector organizations that rely on centralized identity authentication.

The latest technique deliberately inserts invalid parameters into OAuth requests and exploits redirect logic to send users to malicious websites. Attackers then use the EvilProxy man-in-the-middle framework to intercept login credentials and session cookies, and may even distribute malware. Microsoft has not disclosed the number of affected organizations, the financial losses or the exact discovery date, but has warned relevant organizations to strengthen redirect URL validation and session protections.

Mark Radar|MARK RADAR

If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →

All times are in Taipei time (GMT+8)