“Frontier AI models are not effective at finding security vulnerabilities in RTL and hardware design because they have not been trained on a sufficient amount of RTL data.”