There is a newer version of the record available.

Published March 14, 2026 | Version v1.0
Software Restricted

kevinjohnwilliams/AI_Requirements_Framework: AI Requirements Framework v1.0

Authors/Creators

  • 1. NASA

Description

Initial release of the Safety Critical Labs AI Requirements Framework for Safety-Critical Systems.

Establishes nine requirement areas covering AI-specific failure modes not addressed by existing software engineering standards:

  • AI-1: Data Partitioning & Classification
  • AI-2: Addressing AI Bias
  • AI-3: ML Test Coverage
  • AI-4: Continuous Validation
  • AI-5: Hallucination Prevention
  • AI-6: Out-of-Distribution Detection
  • AI-7: Adversarial Robustness
  • AI-8: Explainability
  • AI-9: Human-AI Teaming

Algorithm-agnostic. Applicable to aerospace, aviation, automotive, medical, and other safety-critical domains.

Developed by Kevin Williams / Safety Critical Labs.

Files

Restricted

The record is publicly accessible, but files are restricted to users with access.

Additional details

Related works

Is supplement to
Software: https://github.com/kevinjohnwilliams/AI_Requirements_Framework/tree/v1.0 (URL)

Software

Repository URL
https://github.com/kevinjohnwilliams/AI_Requirements_Framework