Securing the Future: Navigating AI Vulnerabilities and Evolving Security Practices
summary
Dell documents a year of practical work: an internal AI threat library, concrete poisoning categories like label flipping and backdoor attacks, and a push to extend inventories to an AI SBOM that lists datasets, models, vector DBs and RAG components. The proposal bills machine-readable, cryptographically-signed attestations to establish, report and verify integrity; it maps to Google’s supply-chain controls and an EU Cyber Resilience Act use case and calls for converged machine-readable attestations and “secure X ops” for rapid remediation.
AI-written summary · about
chapters (12)
- 0:00AI security challenges
- 4:38AI threat sources and risks
- 7:06Protecting AI data and components
- 10:45Future AI threat landscape
- 13:33AI vulnerabilities and attack types
- 15:22AI supply chain security
- 17:29Remediating AI vulnerabilities
- 20:20Cyber Resiliency Act use case
- 23:09Secure operations for AI components
- 26:25AI supply chain attestation
- 28:53AI vulnerability response processes
- 30:40Q&A
related talks
37mFundamentals of AI safety and security
20mIoT Security: Past, Present and Future
14mDesigning the open and safe AI future
18mCommit Secure AI: Embedding Security in Code, Pipelines, and Production
24mImplementing Robust AI Security for Enterprise Innovation
12mSecuring AI-Driven Finance
44mWhat Security Teams Need to Know and Do
55mAttacking And Protecting Artificial Intelligence