Governance of Generative Artificial Intelligence for Companies
cs.AI, cs.CY, cs.LG
Submitted: 2024-02-05
Updated: 2026-09-05
Comments: This paper is under submission
License: http://creativecommons.org/licenses/by/4.0/
The gist: Generative Artificial Intelligence (GenAI) like ChatGPT has swiftly entered organizations without adequate governance, posing both opportunities and risks.
Terminology
Abstract
Generative Artificial Intelligence (GenAI) like ChatGPT has swiftly entered organizations without adequate governance, posing both opportunities and risks. Limited research addresses organizational governance from both technical and business perspectives. This gap is particularly relevant for international businesses, where differences in regulation, language, and business environments complicate governance. While multiple frameworks for AI governance exist, this needed diversity is lacking for GenAI. This review paper fills this gap by surveying recent literature to better understand the fundamental characteristics of GenAI and to adapt existing governance frameworks specifically to GenAI. The resulting framework delineates scope, objectives, and governance mechanisms designed to both harness business opportunities and mitigate risks associated with GenAI integration. We theorize a distinctive property of GenAI governance: its scope is endogenous to use. Unlike conventional organizational AI, for which governance is organized around a fixed artifact (e.g., model, intended use), GenAI allows users to reconfigure the artifact (e.g., its behavior and risk). Consequently, the object of governance is not fixed ex ante but is partly constituted through use.
Sources
- LLMAuditor: A Framework for Auditing Large Language Models Using Human-in-the-Loop
- A General Language Assistant as a Laboratory for Alignment
- A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivity
- Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity
- On the Opportunities and Risks of Foundation Models
- Toward Generalizable Evaluation in the LLM Era: A Survey Beyond Benchmarks
- Evaluation of Text Generation: A Survey
- Behavioral Use Licensing for Responsible AI
- Multilingual Jailbreak Challenges in Large Language Models
- Safeguarding Large Language Models: A Survey
- Mapping the Ethics of Generative AI: A Comprehensive Scoping Review
- Measuring Massive Multitask Language Understanding
- Ethical Concerns of Generative AI and Mitigation Strategies: A Systematic Mapping Study
- Generative AI and LLMs in Industry: A text-mining Analysis and Critical Evaluation of Guidelines and Policy Statements Across Fourteen Industrial Sectors
- Large Language Models are Zero-Shot Reasoners
- Deduplicating Training Data Makes Language Models Better
- Rethinking Data Protection in the (Generative) Artificial Intelligence Era
- Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models
- The Data Provenance Initiative: A Large Scale Audit of Dataset Licensing & Attribution in AI
- Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection