AegisAI Launches Proteus for Enhanced Email Security

• •Proteus detects over 99% of phishing attacks without relying on known patterns.
• •The model returns verdicts up to 7.2 times faster than leading general-purpose AI models.
• •AegisAI's team developed Proteus to counter the rise of sophisticated AI-generated phishing.

AegisAI has launched Proteus, a foundational model designed to enhance email security by investigating every incoming email like a security analyst. Unlike traditional systems that rely on known patterns, Proteus evaluates emails based on evidence that the sender cannot control, such as link destinations and sender-recipient correspondence history. In a benchmark of 3,020 human-reviewed phishing attacks, Proteus achieved over 99% detection accuracy, significantly outperforming leading general-purpose models, which detected only 81% and 80% of attacks, respectively. The model also returned verdicts up to 7.2 times faster than its competitors. This launch addresses the growing sophistication of AI-generated phishing attacks, which have become increasingly difficult for conventional systems to detect. AegisAI's team, which previously worked at Google, identified the need for a more specialized model to counter the evolving threat landscape. Proteus aims to close the gap in email security by autonomously investigating every message before it reaches the inbox.