Diagrid Dapr 1.18: Verifiable AI Execution & Security

Diagrid has released Dapr 1.18, an update that allows organizations to cryptographically prove how an artificial intelligence agent or workflow executed. This update helps verify who had custody of the work and if its history was altered. The release introduces "verifiable execution" through three new features. Workflow History Signing creates tamper-evident execution records. Workflow History Propagation tracks execution lineage across different systems. And Workflow Attestation passes verified execution context for policy and compliance decisions. These features aim to provide security and compliance teams with a clear chain of custody for decisions made by autonomous systems. Diagrid, a primary contributor to Dapr, designed these features in collaboration with the Dapr community. The update also includes other changes. The Jobs API is now stable, and hot reloading for components and configurations is generally available. Actor applications can now use a single bidirectional gRPC stream, reducing the attack surface. This matters because as AI agents move into production, organizations need to prove what happened in a tamper-proof way, especially when AI approves transactions or accesses sensitive data.