Traditional supply chain risks and AI

In this section, we will look at how traditional supply-chain risks from application development apply to AI. Compared to traditional software development, these risks increase with AI because ML has access to live data. We will focus on techniques that address the risks of an attacker exploiting components in environments with access to sensitive data.

Risks from outdated and vulnerable components

In the complex web of software development, using third-party components has become standard practice. While these components expedite development and reduce costs, they can also introduce a range of vulnerabilities if they are not managed carefully. For instance, outdated libraries and vulnerable frameworks can expose the entire system to various risks, including unauthorized data access, system malfunction, and even legal consequences.

One of the most recent and noteworthy examples in this context is the data breach at OpenAI, a leading organization...