LAAF: A Layered Accountability Architecture Framework for LLM Applications
What it is
This paper presents a systematic review of 122 primary studies and 12 regulatory documents to define a sociotechnical account of accountability for LLMs. It synthesizes these findings into LAAF, an integrated architecture spanning provenance, application logic, human oversight, and governance, mapped to regulations like the EU AI Act and NIST AI RMF.
The problem it solves
The lack of clear mechanisms to trace, explain, and act upon responsibility when LLM outputs contribute to harm in high-stakes sectors, due to the absence of shared accountability metrics and disciplinary disconnection.
What you could use it for
- Designing compliance architectures for high-risk LLM applications under the EU AI Act
- Establishing traceability and redress mechanisms for LLM deployments in healthcare or finance
- Structuring organizational governance and human oversight policies for generative AI systems
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.