GENXAI A RESPONSIBLE AI GOVERNANCE FRAMEWORK FOR GENERATIVE MODELS IN REGULATED INDUSTRIES

The rapid adoption of generative AI in regulated industries as a result of the media coverage it receives has also shown there are very large holes and gaps in our systems and processes for governing, policing and auditing its use.Classic generative models are a well-known class of machine learning models that, however, still have two strange properties: they base the output behavior on probabilities and it is generally hard to understand the generated content itself.We present GENXAI, a responsible AI governance lifeycle framework for generative model development, deployment and operation in regulated industries.GENXAI leverages a life-cycle governance model to embed compliance, ethical risk reduction and explainability into the generative systems workflows.It does this by marrying the following: Data lineage driven governance, Model transparency and human in the loop validation, Iterative monitoring and audit based generation of auditable documents along to keep constantly compliant.One of the most relevant contributions of this work is its centralized governance model, which maps the abstract regulatory and ethical principles into the technical and organizational mandates.GENXAI is applied with analytical comparisons as well as domain-specific use cases on healthcare, the financial services and legal systems. GENXAI A Responsible AI Governance Framework for Generative Models in Regulated Industries