Webinar Description
Key Takeaways
- Addresses security and compliance gaps when deploying open-weight AI models from repositories such as Hugging Face and ModelScope
- Covers detection, inventory and integration of AI models into SBOM and policy-as-code workflows
- Examines AI-specific risks including jailbreaks, prompt injection, data exfiltration and toxic payloads
- Designed for security, compliance and DevSecOps professionals in regulated industries
About the Event
This Anchore webinar focuses on managing security risks associated with open-weight AI models in production environments. As organisations increasingly adopt AI models from public repositories, operational and security teams face challenges in tracking and governing these assets. The session addresses gaps in current software supply chain management practices, arguing that AI models require the same governance rigour applied to open source software components.
Software Supply Chain and AI Model Governance
Traditional Software Composition Analysis (SCA) tools were not designed to handle AI models, leaving organisations without adequate visibility into their AI assets. The webinar explores how to detect and inventory AI models, including their metadata and identifiers, and how to integrate these assets into existing SBOM workflows. Attendees will learn about policy-as-code approaches for enforcing governance requirements across AI model deployments.
AI-Specific Security Risks
The session covers evaluation techniques for risks unique to AI models. These include jailbreaks, prompt injection vulnerabilities, data exfiltration, toxic payloads and import hallucinations. Understanding these attack vectors is essential for organisations seeking to prevent untracked AI models from introducing vulnerabilities or compliance failures.
Who Should Attend
The webinar is suited to security and compliance professionals, DevSecOps teams, software supply chain managers, and technical architects. It is particularly relevant for organisations deploying AI models in production within regulated industries where compliance requirements for AI model tracking are evolving.