SecurityOnlineDo Son
Microsoft Disrupts AI Phishing Service Behind 12,000 Breaches
Part of the Storm-2992 story · since 22 Sep
- Tue 22 Sep, 16:13 UTC. Microsoft, first report. Microsoft Disrupts EvilTokens Phishing Platform After 12,000 Compromises.
- Tue 22 Sep, 20:23 UTC. SecurityOnline, its first report on the story. Microsoft Disrupts AI Phishing Service Behind 12,000 Breaches. This article, 2nd of 3.
- Wed 23 Sep, 11:38 UTC. Security Affairs, its first report on the story. Microsoft Dismantles EvilTokens Phishing Service Behind 12,000 Breaches.
Microsoft and law-enforcement partners have disrupted EvilTokens, a phishing-as-a-service platform that used device-code phishing and artificial intelligence to compromise organisational email accounts. Microsoft said the platform, launched in February 2026, tricked victims into entering attacker-generated temporary codes into legitimate Microsoft sign-in pages.
This completed the normal authentication process and issued valid OAuth access and refresh tokens, allowing criminals to access mailboxes without obtaining passwords.
The service’s AI chatbot analysed email histories to identify financial approval conversations, accounting staff and pending invoices, then helped operators draft requests for urgent wire transfers. Attackers also created hidden inbox rules and, in some cases, registered rogue devices to maintain access. Microsoft linked EvilTokens to more than 12,000 compromised inboxes across over 10,000 organisations, including healthcare, education, finance, construction and wholesale businesses.
The platform reportedly cost $1,500 initially, followed by $500 monthly fees, and used cloud services including Cloudflare Workers, AWS Lambda and Vercel to conceal phishing infrastructure.
Microsoft Threat Intelligence attributes the platform’s development and primary operation to Storm-2992 with moderate confidence. The Metropolitan Police Service arrested two men, aged 32 and 38, in the UK on 11 September 2026; both were released on police bail. Following a US federal court order, Microsoft seized 50 active websites and disabled more than 150 supporting domains, working with Cloudflare, OpenAI and other partners.
Microsoft advises restricting device-code authentication where unnecessary, applying conditional-access controls, revoking active session tokens after compromise and independently verifying payment changes.
How the story developed
3 articles from 3 outlets since 22 Sep. Quiet since 23 Sep. This was the 2nd report, 4 hours after the first.
- Tue 22 Sep, 16:13 UTC. Microsoft, first report. Microsoft Disrupts EvilTokens Phishing Platform After 12,000 Compromises.
- Tue 22 Sep, 20:23 UTC. SecurityOnline, its first report on the story. Microsoft Disrupts AI Phishing Service Behind 12,000 Breaches. This article, 2nd of 3.
- Wed 23 Sep, 11:38 UTC. Security Affairs, its first report on the story. Microsoft Dismantles EvilTokens Phishing Service Behind 12,000 Breaches.
- No CVE
- Larger: an outlet’s first report
- Coverage over time
- Quiet time, folded
Show all 3 articles as a listHide the list
- Tue 22 SepDay 02
- Microsoft Disrupts EvilTokens Phishing Platform After 12,000 Compromises16:13 · Microsoft · First report
- Microsoft Disrupts AI Phishing Service Behind 12,000 Breaches20:23 · SecurityOnline · This article, 2nd of 3
- Wed 23 SepDay 11
- Microsoft Dismantles EvilTokens Phishing Service Behind 12,000 Breaches11:38 · Security Affairs