Tactics for Mitigating Agentic AI Risks

Tech companies are racing to build and sell agentic. The vision is one in which countless AI agents are acting on our behalf: searching the web, making transactions, interacting with other AI agents. But my guest Avijit Ghosh, Applied Policy Researcher at Hugging Face, explains why we’re not even close to having the appropriate safeguards in place. What are the massive gaps and what would it take to close them? That’s the topic of our discussion.

AI Agent Framework: SmolAgents: https://huggingface.co/blog/smolagents

AI Agents Course: https://huggingface.co/learn/agents-course/en/unit0/introduction

Position paper: https://huggingface.co/papers/2502.02649

Op Ed! : https://www.technologyreview.com/2025/03/24/1113647/why-handing-over-total-control-to-ai-agents-would-be-a-huge-mistake/