en.Wedoany.com Reported - U.S. AI cybersecurity company Reken has officially launched its on-device AI security software platform, Reken Private Core, along with its first enterprise security product, Northstar, built on this platform. The platform primarily targets email, instant messaging, and other online communication scenarios. By running AI models locally on user devices, it identifies social engineering attacks, deepfakes, business email fraud, and automated malicious behaviors.
U.S. Reken began developing Reken Private Core in 2024. The core design of the platform is to deploy threat detection capabilities to user endpoints, rather than sending all communication content to external cloud servers for analysis. Messages entering the device are assessed in real-time by proprietary AI models to identify potentially deceptive content, identity impersonation, abnormal interactions, and malicious guidance, reducing the pressure on users to rely on their own experience to identify cyber fraud.
In addition to analyzing communication content, Reken Private Core is equipped with next-generation telemetry sensors to observe abnormal behavioral characteristics within communication channels, identifying AI bots, automated attack programs, and potentially compromised accounts. By combining content analysis with behavioral monitoring, the system not only determines whether a message is suspicious but also continuously observes whether the message sending method, interaction frequency, and account activity are abnormal.
All communication data on the platform is processed on the device to minimize the transmission of sensitive information to external servers. Reken states that its system does not require dedicated GPUs and can run directly on general-purpose hardware, nor does it incur additional token usage costs from invoking external generative AI models. This deployment approach helps enterprises install security capabilities on more employee devices and reduces reliance on centralized high-performance computing resources.
Devices deployed based on Private Core can automatically form the Reken Network, creating a continuous security detection layer for internal enterprise communications and external messages entering the organization. Communications between internal users can be protected through unified policies, while messages from customers, suppliers, and other external accounts automatically undergo analysis for malicious behavior, identity deception, and content manipulation.
Northstar is the first application launched on this platform for enterprise employees, focusing on preventing social engineering attacks, deepfakes, business email compromise, and other AI-driven fraud. The system automatically completes risk identification when employees receive messages and provides security tips to users, shifting threat assessment from post-incident training and manual verification to real-time detection during the communication process.
U.S. Reken believes that traditional security training requires employees to identify fake identities, abnormal links, and fraudulent instructions on their own. However, as the quality of AI-generated content improves, it is increasingly difficult for the human eye to detect anomalies. Northstar is designed to embed detection capabilities directly into the communication environment employees use daily, completing analysis when messages arrive on the device, rather than waiting until employees click on links or execute instructions to address risks.
Northstar is currently in early access. Subsequent deployment priorities will include enterprise device integration, communication channel adaptation, security policy configuration, and operational validation of Private Core across different general-purpose hardware environments.
