SentinelOne cuts malware analysis from weeks to hours with OpenAI Daybreak

Malware reverse engineering is labor-intensive: analysts disassemble binaries, trace behavior and write detections. OpenAI's case study says Daybreak automates much of that triage and analysis, so SentinelOne's team can investigate more threats in the same time.

The story fits a broader pattern of labs courting defenders. Greg Brockman this week highlighted GPT-5.6 Sol finding a critical full-chain Chrome vulnerability in six days and $40,000 of compute, earning Google's first full-chain reward of 2026. Google initially limited Gemini 4 Argon to cybersecurity defenders before general availability, and Mistral tuned Large 4 for cyber tasks. Security is becoming a key battleground for frontier models.

OpenAI also profiled Amp, which runs agents beyond coding on GPT-6 Astra, Ultrafast and GPT-Live-1, showing its newer lineup in production.

Caveats: vendor case studies cherry-pick results, and the same capabilities lower the cost of offense. Watch for independent measurement of detection quality and false-positive rates.