Microsoft traces 1M AI-assisted fraud emails

Microsoft found more than 1 million fraud emails sent between August 3 and 5, with templates showing signs of generative-AI assistance and fake executive approvals for ACH payments of nearly $50,000. Its incident report says 87.7% targeted U.S. recipients and that the messages impersonated CEOs and ServiceNow without evidence either legitimate company was compromised. Finance teams should treat polished, personalized email as a payment-control problem, not a quality signal; Microsoft’s model-routing security analysis shows why cheap detection and human approval still belong in the same loop.