Navigating the Intersection: AI’s Role in Shaping the Secure Open Source Software Ecos...
summary
OpenSSF outlines concrete defensive steps and active problems: OSS can backdoor AI (example hypothetical PyTorch backdoor altering protein design), data poisoning remains real, and LLMs can flood repos with bad fixes. Tools deployed include sigstore for signing models and artifacts and Scorecard scans ~1M repos biweekly to rate supply-chain risk; SLSA (referred to as salsa) defines producer levels. DARPA runs an AI Cyber Challenge with ~$18M prizes using LLMs to find and fix bugs, and the foundation publishes a model signing spec and vulnerability disclosure guidance.
AI-written summary · about
chapters (9)
- 0:00AI and open source security
- 1:49Risks of AI open source packages
- 3:12AI in supply chain
- 5:20Securing AI environments
- 7:48AI and open source risks
- 9:57Secure AI development tools
- 13:49Community initiatives and future
- 15:14OpenSSF projects
- 16:26Q&A
related talks
17mLightning Talk: Navigating the Intersection: AI’s Role in Shaping the Secure Open Sour...
05mIgnite: Security of Open-Source AI: is there any difference?
14mDesigning the open and safe AI future
01h 02mPanel Discussion:The Present and Future of AI and Security
53mCracking the Code: Unveiling Synergies Between Open Source Security and AI
13mSecuring the Software Commons: Standards, Automation, and AI for a Resilie...
50mBuilding AI Security In: MLSecOps in Practice
31mSecuring the Future: Navigating AI Vulnerabilities and Evolving Security Practices