GenAI-Driven Threat Detection with Microsoft Security Copilot

TL;DR

This paper presents DTDA, an adaptive AI agent integrated with Microsoft Security Copilot, that continuously investigates security incidents, uncovers hidden threats, and improves detection precision at industry scale.

Contribution

Introduction of DTDA, a novel autonomous threat detection agent that combines unified data, advanced LLM prompts, hypothesis generation, and dynamic alerting within Microsoft Defender.

Findings

Achieves 80.1% precision in real-world deployment

Recovers hidden malicious activity with 0.78 F1 offline

Processes investigations in median 28 minutes at low cost

Abstract

Defending against today's increasingly sophisticated cyberattacks requires security analysts to continuously translate evolving attacker tradecraft into detection logic. This places defenders in a reactive posture, requiring constantly updated expertise across an increasingly fragmented security landscape. We introduce the Dynamic Threat Detection Agent (DTDA), an always-on adaptive agent that continuously investigates security incidents across Microsoft Defender to uncover hidden threats and generate explainable detections when attack-story gaps are found. DTDA combines: (1) a unified activity timeline spanning alerts, events, user and entity behavior analytics, and threat intelligence; (2) versioned LLM prompt contracts with schema validation, grounding requirements, bounded retries, and fail-closed suppression; (3) a planner-executor investigation loop that generates attack-specific hypotheses and gathers supporting and refuting evidence; and (4) dynamic alert generation with a context-relevant title, severity, MITRE mappings, remediation guidance, implicated entities, and natural-language attack description. Integrated into Microsoft Security Copilot and deployed across tens of thousands of Defender customers, DTDA operates continuously at industry scale. In a 120-day online evaluation, DTDA achieves 80.1% precision from customer feedback while generating novel alerts for approximately 15% of investigated incidents. In offline evaluation, DTDA recovers hidden malicious activity with 0.78 F1 using GPT-5.4, improving over GPT-4.1 by 0.12 F1 and outperforming the baseline by 0.26 F1 points. Operationally, DTDA processes single-incident investigations end-to-end in a median of 28 minutes at a median token cost of USD 2.04, with a 0.38% job-level failure rate. These results demonstrate that autonomous agents can identify missed malicious activity at a production scale.