📙 Prompt, Commit, Repeat: Security at Scale When 1,000 Devs Go AI-Native
summary
Enterprises with thousands of developers face concrete failures: an AI agent nuked a Mac by deleting a home directory, prompt and code injections cause dependency squatting and secrets exfiltration, and agents can push code into repos or CI/CD. The recommended stack enforces an MVP security gate: SSO and MFA, audit logging, sandboxed execution, package proxies, model pinning and cost limits; examples include manage settings (Claude) hooks and human approval for risky ops. Policy treats AI output as third‑party code, requires provenance tracking, continuous validation and incident runbooks.
AI-written summary · about
chapters (6)
- 0:00Introduction and problem setup
- 4:11AI coding tools overview
- 15:33Enterprise risks from AI tools
- 20:48Security requirements and guardrails
- 36:41Implementation and enforcement strategies
- 38:26Q&A session
related talks
23mVibes Not Vulns: Securing the Era of AI-Written Software
35mManaging the AI Surge: Governing & Scaling in the AI Era
19mKeynote: Towards Building Safe and Secure Agentic AI
24mAI Governance for developers: A practical framework for deploying autonomous coding agents safely…
51mBeyond Code Generation: Governing AI Agents Across the DevSecOps Lifecycle
36mHow to Safely Deploy AI Agents that Write and Manage your IaC
01h 03mBuilding Secure Infrastructure for Productive AI Agents
24mHow to Build Trustworthy AI