A survey of practitioners drove key areas of focus for the vulnerability list for AI models: data privacy, safety and bias, and mapping of vulnerabilities.
With AI now impacting every organization, it's time to get a handle on security. NCSA and CISA suggest starting with Secure by Design. Don't stop there.
Supply-chain Levels for Software Artifacts (SLSA) and Sigstore are a good first step toward protecting ML models from attack. But they're not a panacea.
In addition to the extensive list of components in today's software, AI relies on open source AI models and training data. What could possibly go wrong?
Get the best of RL Blog delivered to your in-box weekly to stay up to date on key trends, analysis and best practices across threat intelligence and software supply chain security.
