arXiv:2606.04619v1 Announce Type: new Abstract: We propose MONIR, a Modalized-Output Normative Intermediate Representation for ASP-based compliance reasoning. Its core fragment has a staged operational semantics, while MONIR-ASP provides an executable compilation and extensions for external functions, temporal rules, and stable-model reasoning. We instantiate the framework on Chinese ADAS regulations and standards with an LLM-assisted pipeline. Experiments evaluate extraction quality and the efficiency of modular and incremental ASP solving.
The increasing complexity of AI systems, particularly in critical applications like autonomous vehicles, necessitates robust compliance and explainability frameworks to meet regulatory demands.
This work introduces a structured method for AI systems to interpret and comply with regulations, which is crucial for their deployment in sensitive sectors and for building public trust.
The development of a normative intermediate representation and its application to specific regulations (Chinese ADAS) provides a tangible path toward verifiable AI compliance.
- · AI developers
- · Regulatory bodies
- · Autonomous vehicle industry
- · Compliance software providers
- · Companies ignoring compliance
- · Manual compliance auditors
More AI systems can be demonstrably compliant with relevant regulations, accelerating adoption in regulated industries.
The ability to formally verify AI compliance could lead to new standards for AI auditing and certification.
This could enable the creation of 'regulatory-aware' AI agents that dynamically adapt to legal frameworks in real-time.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI