Creating Active Feedback Loops for Detecting Advanced Attacks - Abnormal Security

Creating Active Feedback Loops for Detecting Advanced Attacks

Abnormal Security prides itself on its differentiated technology and superior efficacy when it comes to stopping advanced email attacks. Despite the overwhelming effectiveness of our platform, like all advanced AI systems we may occasionally miss attacks or incorrectly flag safe messages. In line with our value of customer centricity, our objective is to make the customer experience for reporting these messages as seamless as possible. Additionally, per our value of velocity, we strive to make rapid changes to the system and allow our detection engine to benefit from more data, thereby creating a positive feedback loop that further drives detection efficacy in the future (see here for an example of the value of data in machine learning systems).

To better align with these objectives, we have revamped our user experience for reporting incorrectly flagged messages. Our new Detection 360° tab in the Portal navigation bar provides a consolidated view where one can report messages and view prior reports and status updates. In addition to this redesign for increased user friendliness, we also have new cards with general detection statistics and the ability to filter reports on status and date which adds convenience and visibility into the consistent improvements of our AI engine. 

We hope this added functionality provides confidence in our AI systems and the degree to which we use customer feedback to ensure we are always getting better and increasing performance. Such improvements are critical as we continue our quest to detect and remediate every advanced, harmful email attack that targets our customers’ environments.