Understand this story
Uber Engineering shared its internal agent security system, Secure AI, whose design not only checks agent permissions but also verifies whether specific actions match the user's request. The system consists of three parts: pre-launch review, runtime constraints, and continuous testing, but it did not disclose quantitative protection results. Before launch, the AI Registry registers the agent's purpose, owner, allowed behaviors, and risk level, while Argus analyzes the code, prompts, and tool invocation paths; at runtime, AI Guard makes judgments by combining the user request, role, session context, and agent purpose, and uses small local specialized models to run checks in parallel to control latency. The team adopted a shadow mode that observes without blocking first and calibrates agent by agent, while acknowledging that protection can only reduce risk and that effectiveness against novel prompt injection remains relatively hard to predict.
AI-generated from public reports
Latest development · Report
Uber Shares Agent Security Practice: Secure AI's Three-Layer Defense DesignDevelopments
1 confirmed developments
- Source:Shao Meng
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 7
Shao Meng
Uber Shares Agent Security Practice: Secure AI's Three-Layer Defense Design
Uber Engineering shared its internal agent security system, Secure AI. In addition to checking agent permissions, it also verifies whether specific actions match the user's request. It consists of three parts: pre-launch review, runtime constraints, and continuous testing, but it did not disclose quantitative protection results. Before launch, the AI Registry records the agent's purpose, owner, allowed behaviors, and risk level, while Argus analyzes the code, prompts, and tool-calling paths; at runtime, AI Guard makes judgments based on the user's request, role, session context, and agent purpose, and uses small local specialized models to run checks in parallel to control latency. The team adopted a shadow mode that observes without blocking first and calibrates agent by agent, while acknowledging that protection can only reduce risk and that effectiveness against novel prompt injection remains difficult to predict.