Evidence search
Search
Search titles, summaries, topics, providers, authors, and the full open-access corpus.
Results for “Trust boundaries”
9 publicationsDesign AI agents that contain indirect prompt injection
Filters lower the odds that an agent obeys injected text. Architecture decides what an obeying agent can reach. Compare six published patterns and CaMeL by what each removes, what it costs and what it still misses.
AI systems · Google DeepMind / AgentDojo / NIST / OWASP / Microsoft · By Cloud Security DeskGive Kubernetes pods cloud credentials without static keys
EKS Pod Identity, IRSA, AKS Workload ID and Workload Identity Federation for GKE all swap a projected token for short-lived credentials. What leaks if you stop there is the node's own identity.
Identity & access · Amazon Web Services / Microsoft Azure / Google Cloud / Kubernetes · By Cloud Security DeskGive Azure readers access to one resource group
Assign Reader without granting write access or assuming it grants data access.
Identity & access · Azure · By Cloud Security DeskSet explicit trust boundaries for Entra partner access
Accepting another tenant’s authentication claims is a specific trust decision, not blanket approval of its users, devices or access to your applications.
Identity & access · Microsoft · By Cloud Security DeskSecurity evidence for AI release decisions
A release approval should identify the changed application, the claims its tests support and the evidence that expires when a model, prompt, data path or runtime changes.
AI systems · By Cloud Security DeskWhere fine tuning data needs a trust boundary
A training dataset can preserve its checksum and still teach the wrong behavior. Admission controls need to separate origin, transformation, approved use and the model change they produce.
AI systems · By Cloud Security DeskThe telemetry collector is part of the evidence boundary
Review sender identity, tenant routing, processing and export as separate trust boundaries before treating collected telemetry as dependable evidence.
Detection & response · OpenTelemetry · By Cloud Security DeskModel output needs its own trust boundary
A model can produce valid JSON containing an unauthorized identifier, an unsafe link or text that a renderer interprets as code. The application consuming that output owns the next trust decision.
AI systems · OpenAI · By Cloud Security DeskQwen3.8-Flash-Next and GLM-5.3-Flash share a 3:1 long-context pattern
Both models replace most conventional attention layers with recurrent state and reserve sparse attention for periodic retrieval. Their differences lie in where they place capacity, how much neural computation they activate, and what their serving stacks must keep trustworthy.
AI systems · Resilience · By Umair Akbar and Ahmed Elshekh