Bird
0
0

How can you combine sandboxing with logging to improve safety when running unknown AI code?

hard📝 Application Q9 of 15
Agentic AI - Agent Safety and Guardrails
How can you combine sandboxing with logging to improve safety when running unknown AI code?
ARun code normally and log all outputs
BRun code without sandbox and log errors only
CRun code in sandbox without logging
DRun code in sandbox and log all system calls made
Step-by-Step Solution
Solution:

  1. Step 1: Understand sandbox and logging roles

    Sandbox isolates code; logging tracks actions for review.

  2. Step 2: Combine for safety

    Logging all system calls inside sandbox helps detect suspicious behavior safely.

  3. Final Answer:

    Run code in sandbox and log all system calls made -> Option D

  4. Quick Check:

    Sandbox + logging system calls = safer monitoring [OK]
Quick Trick: Log system calls inside sandbox to monitor unknown code safely [OK]
Common Mistakes:
  • Logging only errors misses suspicious actions
  • Running without sandbox risks system
  • Ignoring logging reduces safety

Want More Practice?

15+ quiz questions · All difficulty levels · Free

Free Signup - Practice All Questions
More Agentic AI Quizzes
Agent Observability - Token usage and cost tracking - Quiz 4medium Agent Observability - Latency monitoring per step - Quiz 13medium Agent Safety and Guardrails - Rate limiting and budget controls - Quiz 2easy Agent Safety and Guardrails - Rate limiting and budget controls - Quiz 15hard Future of AI Agents - Self-improving agents - Quiz 14medium Production Agent Architecture - Agent API design patterns - Quiz 11easy Production Agent Architecture - Async agent execution - Quiz 15hard Real-World Agent Applications - Data analysis agent pipeline - Quiz 2easy Real-World Agent Applications - Personal assistant agent patterns - Quiz 8hard Real-World Agent Applications - Customer support agent architecture - Quiz 11easy