Uber Open-Sources ADR: An Enterprise Security Framework for AI Agents
brucemacv · x · 2026-08-03
Uber has open-sourced ADR (Agentic AI Detection and Response), an enterprise-grade security framework designed to monitor and mitigate security risks posed by AI agents. Already deployed in production at Uber, the accompanying research paper was accepted by MLSys 2026.
Key features include:
- Observability & Telemetry: Collects AI agent behavioral data for analysis.
- Security Benchmarking: Includes a suite of over 300 security tasks and supports 133 MCP servers.
- Threat Detection: Identifies risky behaviors in enterprise tools like Cursor, Claude Code, and customer-facing agents.
- Cross-Platform: Supports macOS, Linux, and Windows with reproducible benchmarking workflows.
Related event: Uber Open-Sources Agentic AI Security Framework ADR(2 posts)→
More from coding & agent
- Emergent Behavior: Hermes Autonomously Steps in to Correct Claude — jasonkneen · 2026-08-03
- Kimi K3 Tops pmpp-hard Eval: Open Model Achieves SOTA in Hardcore Agent Tasks — xeophon · 2026-08-03
- Ditch Xcode: Rork Max Lets You Build and Preview Native Apps Directly on iPhone — rudrank · 2026-08-03
- AI Becomes a Lifesaver for Resolving Complex Merge Conflicts — haltakov · 2026-08-03
- Turn Your Obsidian Notes Into Kanban Boards Natively with Bases — dSebastien · 2026-08-03
- Eve framework: batteries-included AI agent deployment with durable execution and sandbox — evilrabbit_ · 2026-08-03