Legal Regimes Governing the Use of Prompts (Prompt Engineering) in the Training and Deployment of Large Language Models

Abstract

The paper examines the legal regimes governing the use of prompts in the training and deployment of large language models (LLMs) and generative artificial intelligence systems (AI systems). It analyzes risks, including prompt injection that may result in unauthorized access to information; intellectual property rights infringements, and AI systems manipulation. The paper further addresses the protection of prompts as objects of intellectual property, including protection under copyright law and trade secret regimes. Particular attention is devoted to jurisprudence in various jurisdictions, including China and the United States, where courts are developing approaches to determining the extent of human creative contribution in works generated with the assistance of AI. The study emphasizes the need for regulatory and technical measures to ensure security, oversight, and the allocation of responsibility in the use of generative AI systems.