arXiv: HTTP REST API Structure Learning

This paper, published on arXiv, introduces a new technical framework for learning the structure of causal relationships within REST APIs, specifically designed to support AI safety compliance. It proposes a method for automatically mapping how data flows and decisions are made across API endpoints, which is critical for auditing and verifying that AI systems operate as intended. While not a regulatory mandate itself, this publication provides a practical tool for implementing transparency and explainability requirements under emerging AI regulations.

Organizations deploying AI systems that rely on complex API interactions, particularly in regulated sectors like finance, healthcare, and critical infrastructure, are most affected. Compliance teams in these sectors must now consider how to document and validate the decision-making logic of their AI services, as regulators increasingly expect demonstrable control over model behavior and data lineage.

Compliance teams should immediately assess whether their current API documentation and model governance frameworks capture causal dependencies between inputs and outputs. They should pilot this structure learning approach on high-risk AI systems to generate auditable maps of API logic, and prepare to integrate such technical evidence into their regulatory filings. Proactive adoption will reduce the risk of non-compliance as enforcement of AI safety rules intensifies.