OpenAI said on October 14 that SentinelOne, a US-based cybersecurity company, is using Daybreak, which gives defenders access to OpenAI models for cybersecurity work.

SentinelOne chief AI officer Gregor Stewart said Daybreak can produce human-level malware reverse-engineering output in about eight hours.

He said a top researcher can spend four to six weeks analyzing a sophisticated malware sample.

Stewart said his team remains accountable for the results. He added that decisions, such as architecture changes, should still have human oversight.

OpenAI said faster analysis lets SentinelOne compare more malware samples and identify shared components and links between threats.

Stewart also wants to use OpenAI’s cybersecurity capabilities in code creation so SentinelOne developers can address security issues as they build.

🔗 Source: OpenAI

Recent SentinelOne developments