Evidence · DiggingBeagle record
OpenAI and Block: case study on AI-assisted security work
October 14 first-party deployment case study describes Block using automated vulnerability, investigation and detection workflows with human judgment and approval. Metrics are vendor/customer-reported, not independent savings measurements.
- Published
- Oct 14, 2026
- Publisher
- OpenAI and Block
- Source role
- vendor statement
Evidence record
October 14 first-party deployment case study describes Block using automated vulnerability, investigation and detection workflows with human judgment and approval. Metrics are vendor/customer-reported, not independent savings measurements.
Claim-level citations (2)
- supportsBlock reports AI-assisted patch proposals and accelerated detection replay: The case study reports a proposed patch was generated 18 minutes after a high-level vulnerability submission during recent red-team exercises; it does not say production deployment was approved in 18 minutes.
Results at a glance: proposed patch
- supportsBlock reports AI-assisted patch proposals and accelerated detection replay: The case study reports replaying a proposed detection change against seven months of telemetry in under one hour.
Results at a glance: detection tuning