Microsoft has revealed that it led an industry-wide effort to disrupt a subscription-based scam platform known as EvilTokens. This platform used an AI-powered chatbot to compromise over 12,000 Microsoft accounts within a few months. The operation, which began with the launch of a Telegram channel in February, required users to pay an initial fee of $1,500, followed by a recurring monthly charge of $500. EvilTokens provided a streamlined, end-to-end system that made it easier for cybercriminals to compromise email accounts in bulk, reducing the complexity of traditional phishing and fraud methods. Once access was gained, the platform allowed users to analyze victims' inboxes, identify high-value targets, and draft emails designed to trick employees into transferring money to accounts controlled by attackers. The AI chatbot played a central role in this process, analyzing inbox data to help criminals pinpoint trusted relationships, payment authorizations, and roles that made fraud more likely to succeed. It also suggested specific strategies, such as impersonating colleagues or vendors, to increase the chances of a successful scam. The platform's efficiency made it particularly dangerous, as it reduced the technical barriers for individuals with limited cybercrime experience to conduct large-scale fraud. By providing tools that automated many aspects of the attack, EvilTokens lowered the risk and complexity typically associated with such operations. This made it easier for a broader range of cybercriminals to engage in financial fraud against businesses and individuals. Microsoft's intervention, alongside other industry players, has led to the disruption of EvilTokens, preventing further exploitation of its platform. While the exact methods used to take down the operation have not been disclosed, the effort highlights the growing collaboration among tech companies to combat emerging cyber threats. The incident also underscores the increasing sophistication of scam platforms, which now leverage AI and subscription models to operate more efficiently and at scale.