Definition
An attack that compromises an AI developer's development tools, package repositories (npm, PyPI), CI/CD pipelines, or IDE plugins to inject malicious code, steal credentials, or tamper with AI configurations. Unlike attacks on production systems, supply-chain attacks hit developers before code even reaches deployment.
Why it matters
AI developers are high-value targets: their credentials unlock frontier models, cloud infrastructure, and proprietary training data. A single compromised developer can become the entry point for attackers to compromise entire organizations or inject poisoned code into production AI pipelines.