EvilTokens was an AI-powered phishing service
Assessment
The 'AI-powered' label is Microsoft's own and is accurate in that the platform integrated an AI chatbot as a core, marketed feature. The nuance a reader should keep is what the AI actually did: it analysed already-compromised inboxes — summarising and translating emails, mapping organisational roles, surfacing wire-transfer conversations, and recommending impersonation targets — to accelerate business-email-compromise fraud. The account takeover itself used conventional device-code phishing, not AI. So 'AI-powered phishing' is fair shorthand for the product, but read literally as 'AI conducted the phishing attacks' it overstates the AI's role, which was reconnaissance and targeting after access was already obtained by ordinary means.
Where this claim appeared
CSO Online · 2026-09-23
https://www.csoonline.com/article/4225175/microsofts-eviltokens-takedown-sheds-light-on-state-of-ai-powered-cybercrime.htmlWhat “Assessed, Not Confirmed” means
A named source states this as its own assessment, at its own stated confidence, rather than as established fact. Attribution to a nation state usually sits here. The assessment is real and reportable; treating it as settled is the error.
4 of 5 · rating scale
Assessed in
EvilTokens: Microsoft's DCU dismantles an AI-chatbot phishing service that hit 12,000 inboxesThink this assessment is wrong? Report an error.