The October 14 OpenAI case study quotes SentinelOne’s AI executive: Daybreak reportedly produces human-level malware reverse engineering in about eight hours, compared with four to six weeks for an expert in the described example. The comparison comes from an interview, not a published benchmark protocol.
The account presents automation as a way to investigate more threats. It does not demonstrate absence of errors or equivalence across all deliverables and malware families. The dated event is publication of the customer case.