Supply chain

- [Narrator] If you or someone close to you has a life-threatening food allergy, then you understand the importance of knowing the ingredients in the food you buy. AI models and the software used to run AI workflows are similar. Using pretrained or foundation models can help jumpstart your next AI project, will delivering time and cost savings compared to starting from scratch. And downloading a public dataset for training could save months of having to collect your own data. However, using existing data and components is not without risk. Models from untrusted sources can contain malicious code and back doors. Public datasets may contain data you don't want your model trained on, and it can be hard to analyze third party models and software for accuracy and resilience. And new versions of software in the AI supply chain can introduce vulnerabilities. Increasingly, companies are using SBOMs, or Software Bills of…