Scalable Security and Compliance in the Age of AI
summary
Findings from 37,000 AI dependency recommendations and a review of 200,000 malicious OSS packages show more than a quarter of AI suggestions are wrong and registries now host hundreds of thousands of malicious uploads (454,000 in 2025) with 43,000 remaining downloadable for years. Recommended actions include using deterministic scanners (Privateier, Sonatype patterns, Dependabot gaps), machine-readable controls (Jamara/Jamara specs, LFX insights), and tying OpenSSF/Finnos/CNCF guidance into a standardized, automated compliance pipeline.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:46AI security threats
- 9:11Proactive defense
- 15:29Governance lagging compliance
- 20:51Standardizing security workflows
- 23:20Demo: security dashboards
- 26:08Steps to scalable security
- 27:52Q&A
related talks
33mSecuring the AI Surge
21mNavigating the Intersection: AI’s Role in Shaping the Secure Open Source Software Ecos...
28mOwn Goal: How AI Agents Are Quietly Scoring for the Other Side
31mSecuring the Future: Navigating AI Vulnerabilities and Evolving Security Practices
44mThe New Normal: Collaborative Strategies in a Regulated, Secure, and AI World
05mIgnite: Security of Open-Source AI: is there any difference?
35mManaging the AI Surge: Governing & Scaling in the AI Era
22mThe AI Accountability Mandate: Securing the Software Supply Chain