The illusion of intelligence: How AI washing shapes startup narratives and venture capital decisions

This Master’s final project investigates the phenomenon of AI washing within the venture capital ecosystem. It studies how AI-related claims made by startups are associated with actual underlying technological capabilities, and analyzes the funding outcomes of such companies. AI washing is known as the strategic exaggeration of a company’s use of artificial intelligence with the goal of appearing more innovative and attractive to investors, despite providing limited evidence of underlying technological capability. Since in most recent years artificial intelligence has become one of the dominant and fastest-growing technological narratives in financial markets, venture capital investors increasingly allocate capital towards firms positioned as “AI-driven”. Nonetheless, the complexity of AI technologies can cause substantial information asymmetries that make it difficult for investors to distinguish between genuine technological advancement and strategic but symbolic positioning. The thesis combines insights from signaling theory, information asymmetry, and legitimacy theory to explain how and why AI washing emerges in the startup ecosystem. In particular, the analysis examines whether AI-related claims function as effective signals of technological capability, or if they function as merely low-cost marketing strategies. The analysis is based on a sample of 100 venture-backed companies extracted from Crunchbase in the time period between 2019 and 2025. To operationalize AI washing, the thesis develops a structured AI substance framework based on five indicators: evidence of proprietary AI models, evidence of data and model infrastructure, presence of specialized personnel operating in the AI domain, AI-related research activity, and centrality of AI in the value proposition. These indicators are then aggregated into an AI substance score ranging from 0 to 5, from which the AI washing variable is derived. Results show that AI washing is observed in a meaningful proportion of the sample. Approximately 26% of companies exhibit limited observable evidence supporting their claims, and are therefore classified as engaging in AI washing. Simultaneously, the majority of firms analyzed demonstrate moderate or substantial levels of technological substance, unveiling how symbolic narratives coexist genuine technological innovation rather than replacing it. Furthermore, a regression analysis is carried out. Findings show how firms with stronger AI substance are consistently able to obtain greater funding outcomes, even after controlling for firm stage, number of funding rounds, and number of lead investors, suggesting that in the end investors value technological capability over general narratives. Moreover, AI washing has a negative and statistically significant impact on funding outcome, indicating that companies engaging in AI washing differ significantly from companies with genuine AI capabilities in terms of amount raised. Overall, the findings suggest thar although AI-related narratives may help startups gain initial visibility and legitimacy, VC markets keep placing a higher value on credible and verifiable indicators of technological substance.