Threat protection-focused startup Deep Instinct Ltd. today announced the launch of DIANNA — short for Deep Instinct’s Artificial Neural Network Assistant, a new generative artificial intelligence assistant that provides expert-level malware analysis for unknown threats.

Claimed to offer an unprecedented level of insight into unknown threats, DIANNA is powered by a large language model and serves as a virtual AI team of malware analysts and incident response specialists. The service provides deep analysis of known and unknown attacks, revealing the techniques employed and behaviors of files to facilitate understanding and mitigation of threats before a breach.

Deep Instinct argues that current cybersecurity solutions often only leverage generative AI to summarize data from existing sources like logs and reputation engines. Though valuable, the approach offers only retrospective analysis with limited context. The company harnesses generative AI to empower DIANNA with the collective knowledge of cybersecurity experts embedded within the LLM to provide in-depth malware analysis of unknown files and identify malicious intent with accuracy.

“With the rise in AI-generated attacks, organizations can no longer be complacent or reactive in how they approach cybersecurity,” said Chief Executive Lane Bess. “It’s time to fight AI with better AI and raise greater awareness about the unknown threats impacting businesses.”

DIANNA integrates with Deep Instinct’s deep learning-powered prevention-first capabilities to provide in-depth insights into known and unknown attack behavior through static analysis. DIANNA doesn’t just provide the classification results but also provides in-depth analysis and reporting in a clear, digestible way, allowing security teams to make informed decisions and prioritize threats effectively, optimizing security operation center performance, according to the company.

Key features of DIANNA include insights into unknown scripts, documents and binaries, which are crucial for combatting zero-day attacks, those with no patch yet. The service also translates code from various languages into comprehensive natural language reports, not only analyzing code, but also understanding its intent and potential actions.

DIANNA also enhances visibility into Deep Instinct’s prevention model decision-making, helping organizations refine their security measures. By automating the more cumbersome tasks of SOC analysis, DIANNA streamlines workflows, allowing security teams to concentrate on strategic initiatives, the company said.

Deep Instinct is a venture capital-backed startup that has raised $322 million in funding, according to Tracxn, including a round of $43 million in February 2020. Investors include Millennium New Horizons LLC, Unbound Investors Inc., LG Electronics Inc. and Nvidia Corp.

Image: Deep Instinct