Towards Transparent and Compliant AI Marketing with a Dataset and Benchmark for Detecting AI Washing

With the rapid proliferation of artificial intelligence (AI) technologies, the marketing of products and services has seen a rise in exaggerated or misleading claims about AI capabilities, a phenomenon known as “AI washing.” These deceptive practices not only misinform consumers but can also have legal and ethical implications, with regulatory bodies like the U.S. Securities and Exchange Commission (SEC) taking action against companies making false AI-related claims. This paper introduces a novel dataset and benchmark for the detection of AI washing in marketing text. Our dataset consists of a diverse collection of promotional materials, annotated with labels identifying whether the claims constitute AI washing or represent genuine AI use. We provide an evaluation framework that enables the assessment of various AI washing detection methods, advancing the development of tools to identify and address these misleading practices.