Topic starter
12/02/2026 9:44 pm
TL;DR: AI applications now depend on a separate Data and AI Lifecycle with distinct data, notebook, pipeline, registry, and runtime stages, and the source article argues that this creates a security blind spot for AppSec teams, according to Noma Security. The governance gap is widening because these assets behave like non-human identities with persistent access and tool reach that conventional SDLC controls were never built to track.
Editorial analysis by NHI Mgmt Group, based on content published by Noma Security: “Data & ai lifecycle 101: unpacking the unique stages, tools, and technologies”.
Key questions
Q: How should security teams secure the Data & AI lifecycle without treating it like a separate island from application security?
A: Security teams should treat the Data & AI lifecycle as part of the application surface, not as a parallel world.
Q: Why do AI pipelines and model registries create governance risk?
A: AI pipelines and model registries create governance risk because they control what data is used, what model version is deployed, and which jobs can execute automatically.
Q: What should teams do to secure AI runtime behaviour?
A: Teams should define controls for prompts, retrieval sources, model endpoints and telemetry, because runtime is where model behaviour becomes dynamic.
Practitioner guidance
- Map the AI lifecycle separately from the SDLC Document data preparation, training, deployment and runtime as distinct control domains with named owners, hand-offs and approvals.
- Inventory notebooks, pipelines and registries Capture every notebook environment, pipeline system and model registry that can read data, change models or trigger deployment.
- Define runtime governance for prompts and retrieval Set controls for model endpoints, retrieval-augmented generation sources, prompt handling and runtime telemetry so behaviour is observable.
Bottom line: AI applications now depend on a separate lifecycle of data, model and runtime assets that standard SDLC controls do not fully cover.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
This topic was modified 4 days ago by NHI Mgmt Group
08/10/2026 5:42 pm
Data and AI lifecycle security is becoming a separate governance domain, not a subtopic of AppSec. The lifecycle described in the source article has its own assets, workflows and trust boundaries, and those do not map cleanly onto conventional SDLC thinking. That is why application security leaders experience it as a black box: the security problem lives in data movement, model handling and runtime behaviour as much as in code. Practitioners should treat AI lifecycle governance as a distinct operating model.
A few things that frame the scale:
- AI-related credential leaks surged 81.5% year-over-year in 2025, with the surrounding AI infrastructure leaking 5x faster than core LLM providers, according to the State of Secrets Sprawl 2026.
A question worth separating out:
Q: How can organisations tell whether their AI governance evidence is complete?
A: A complete evidence set can answer four questions without reconstruction: what the model received, what it returned, what version and policy were active, and who was accountable for the change or action. If any of those answers depend on Slack, screenshots, or memory, the evidence model is incomplete and should be redesigned.
👉 Read our full editorial: Data and AI lifecycle security is becoming an AppSec blind spot