SentinelOne says OpenAI Daybreak reduced the time needed to reverse-engineer sophisticated malware from four to six weeks to roughly eight hours, giving its security researchers capacity to investigate more samples. Gregor Stewart, SentinelOne’s chief AI officer, says the system produces human-level reverse-engineering output in that time. Reverse engineering means taking software apart to understand how it behaves; researchers often revisit earlier steps as they learn more.

Daybreak gives defenders access to OpenAI models for cybersecurity work, and SentinelOne uses them to examine malware. Stewart says researchers remain accountable for decisions: changes to software architecture still need human review, while point fixes that do not fundamentally change how software works may be suitable for automation. He says faster analysis lets teams examine threats they previously could not afford to investigate deeply.

Comparing more samples can reveal shared components between apparently unrelated malware and give researchers leads in customer investigations. Stewart says seeing broader connections could help teams defend further out and act more proactively. He also says access to OpenAI’s cyber capabilities is spreading security work beyond a small group of specialists, and he wants to bring those capabilities into code creation so developers can address security issues while building software.⟂Daybreak lets SentinelOne use OpenAI models to reverse-engineer malware, reducing analysis from four to six weeks to about eight hours.⟂Gregor Stewart says architecture changes still need human review, while some point fixes may be automated.⟂Comparing more malware samples can reveal shared components and threat relationships, while expanding who can take on security work.

SentinelOne says OpenAI Daybreak reduced the time needed to reverse-engineer sophisticated malware from four to six weeks to roughly eight hours, giving its security researchers capacity to investigate more samples. Gregor Stewart, SentinelOne’s chief AI officer, says the system produces human-level reverse-engineering output in that time. Reverse engineering means taking software apart to understand how it behaves; researchers often revisit earlier steps as they learn more.

Daybreak gives defenders access to OpenAI models for cybersecurity work, and SentinelOne uses them to examine malware. Stewart says researchers remain accountable for decisions: changes to software architecture still need human review, while point fixes that do not fundamentally change how software works may be suitable for automation. He says faster analysis lets teams examine threats they previously could not afford to investigate deeply.

Comparing more samples can reveal shared components between apparently unrelated malware and give researchers leads in customer investigations. Stewart says seeing broader connections could help teams defend further out and act more proactively. He also says access to OpenAI’s cyber capabilities is spreading security work beyond a small group of specialists, and he wants to bring those capabilities into code creation so developers can address security issues while building software.⟂Daybreak lets SentinelOne use OpenAI models to reverse-engineer malware, reducing analysis from four to six weeks to about eight hours.⟂Gregor Stewart says architecture changes still need human review, while some point fixes may be automated.⟂Comparing more malware samples can reveal shared components and threat relationships, while expanding who can take on security work.