NSERC I2I Phase Ia: An Intelligent Framework for Social Engineering Cyber Security Training

人类是网络安全中最薄弱的环节，事实上，比计算机更容易受到攻击。近年来，由于人为失误和缺乏适当的安全培训和意识，无数重大安全事件取得了成功。在这方面，虽然市场上有各种各样的网络安全产品来保护计算机和网络，但几乎没有或很少有不成熟的产品来培养和培训用户的网络安全意识。目前，组织仍然依赖于实践研讨会、在线课程和传统的网络安全意识培训方法，这些方法无效且不可扩展。在这方面，我们为网络安全意识培训开发了一种新的多层技术，并申请了专利，以帮助组织了解和减轻与社会工程威胁相关的潜在安全风险。我们的技术利用人工智能、游戏化理论和网络安全策略的原理，使组织能够实施社会工程防火墙。该系统通过测量组织资产在公共资源（包括Internet、社会网络和媒体等）上的数字足迹，为给定组织提供潜在社会工程威胁的估计。基于估计的社会工程威胁，使用深度强化学习（RL）算法，我们的技术构建并执行了一个交互式反社会工程训练计划，该计划结合了被动和主动训练。然后，我们的技术创建并推荐社会工程防火墙安全策略，该策略用于组织网络中，以减少和减轻任何潜在社会工程威胁的风险。这个项目的主要目标是开发一个可扩展的原型，它提供了一个概念证明，即强化学习方法将制作和执行特定于用户的社会工程攻击。RL原型旨在通过不同类型的媒体（包括社交/专业/研究网络和/或电子邮件）执行社会工程攻击，与用户进行交互。

Human beings are the weakest link in cybersecurity, and are, in fact, more vulnerable than computers. In recent years, countless major security incidents have succeeded because of human errors and the lack of appropriate security training and awareness. In this regard, although there is a wide range of cybersecurity products in the market to protect computers and networks, almost none or a few immature products exist for developing and training users in cybersecurity awareness. At present, organizations are still relying on hands-on workshops, online courses, and traditional training methods for cybersecurity awareness, which are ineffective and unscalable. In this regard, we have developed and patented a new multilayer technology for cybersecurity awareness training to help organizations understand and mitigate potential security risks associated with social engineering threats. Our technology utilizes the principles of artificial intelligence, gamification theory and cybersecurity strategies for enabling organizations to implement a social engineering firewall. The system provides an estimation of potential social engineering threats for a given organization by measuring the digital footprint of the organization assets on public sources, including the Internet, social networks and media, among others. Based on the estimated social engineering threats, using deep reinforcement learning (RL) algorithms our technology constructs and executes an interactive anti-social engineering training program that combines both passive and active training. Our technology then creates and recommends a social engineering firewall security strategy that is used within the organization network to reduce and mitigate the risk of any potential social engineering threats. The main goal of this project is to develop a scalable prototype that gives a proof of concept that the reinforcement learning approach will craft and execute users-specific social engineering attacks. The RL prototype aims at interacting with the users by performing social engineering attacks via different types of media, including social/professional/research networks and/or emails.