Traditional cybersecurity tools use LLMs to act like a super-powered search engine. These LLMs sift through mountains of data, such as logs from your computer and security reports from other sources, to identify patterns and summarize potential threats. They are good at flagging suspicious activity based on what they've seen before.
But here's the limitation: These tools struggle with entirely new attacks, often called "zero-day threats." They can tell you something weird is happening, but they can't explain why it's malicious. It would be as if the LLM was reading a report about a new type of flu. It identifies symptoms and tells you it's different from known illnesses, but it can't explain how the flu makes you sick without additional information.
This retrospective analysis is valuable, but it's reactive. By the time the LLM identifies a threat, it might already be causing damage. What is needed is a system that understands the "how" behind malicious behavior.
Deep Instinct, the prevention-first cybersecurity company that stops unknown malware pre-execution with a purpose-built, AI-based deep learning framework, has a solution to that: Deep Instinct’s Artificial Neural Network Assistant, or DIANNA, an AI-based cybersecurity companion that provides explainability into unknown threats.
DIANNA integrates with Deep Instinct’s existing prevention technology to provide detailed analysis of known and unknown threats through static examination. Unlike traditional machine learning tools, DIANNA goes beyond simple classification. It delivers in-depth analysis and clear reports to allow security teams to make informed decisions and prioritize threats efficiently.
DIANNA also translates code from various languages into clear reports. It doesn't just analyze the code; it deciphers its intent and potential actions, explaining what the code does, why it's malicious and how it might harm systems.
DIANNA provides insights into Deep Instinct's prevention models' decision-making, which allows organizations to refine their security posture for optimal effectiveness. Furthermore, DIANNA analyzes various file formats, including binaries, scripts, documents, shortcuts and other potential threats. It also automates tedious SOC analysis tasks to free up security teams for more strategic initiatives.
“There are two factors that set DIANNA apart from other AI-powered chatbots,” said Yariv Fishman, Chief Product Officer of Deep Instinct. “First, its unprecedented malware analysis compresses hours of work, requiring deep cyber threat expertise, into seconds. Second, DIANNA’s ability to analyze unknown threats, including scripts, documents, and raw binary files, is unmatched. Both capabilities build upon our prevention-first approach and allow security teams to focus on what truly matters.”
The long-story-short of this? DIANNA improves SOC performance and reduces the time to resolve incidents while boosting job satisfaction by minimizing wasted effort on false positives.
“DIANNA provides vital threat explainability, enhances our prevention-first approach, and marks a strategic shift towards a more informed, efficient and effective cybersecurity environment,” said Lane Bess, CEO of Deep Instinct.
Edited by
Alex Passett
