This paper investigates the impact of generative artificial intelligence on workers’ decisions to found businesses in the United States. Although there has been considerable speculation that generative AI can lower barriers to innovation, encourage self-employment, and stimulate new venture formation, systematic empirical evidence on this question remains limited. To address this gap, we merge occupation-level exposure measures to generative AI (as developed by Felten et al., 2023) with individual-level data from the Current Population Survey. We then implement a continuous difference-in-differences design to compare changes in self-employment outcomes among workers who are differentially exposed to generative AI technologies before and after ChatGPT’s public release. Our results indicate that individuals employed in occupations more exposed to generative AI exhibit a reduction in transitions to incorporated self-employment in the period following ChatGPT’s release, in contrast to workers in occupations with lower exposure. The effect on unincorporated self-employment is statistically indistinguishable from zero, and we find no significant change in unemployment rates. While generative AI holds promise for facilitating the search for new ideas and boosting productivity, these findings suggest that technology-induced entrepreneurship may not manifest uniformly in the short run, possibly due to uncertainty, reallocation frictions, or the possibility that workers choose to leverage these tools in existing paid employment roles rather than in newly formed businesses.