← Home
Latent Space · April 25, 2025 · 70m

AI Security: New Attack Surfaces, New Defenses

Alessio Fanelli
SpotifyApple Podcasts

Mapping the emerging AI security landscape: prompt injection, data poisoning, model theft, and the new attack surfaces created by AI agents that act autonomously.

This summary was generated from show notes and public descriptions, not from a full transcript review. Details may contain inaccuracies.

Canon

•
AI agents expand the attack surface — new environments create new vulnerabilities
Every new tool integration and environment access granted to an AI agent creates a new attack surface. The richer the agent's environment, the larger the security risk.
ClaudeChatGPTGemini
•
Falsehood in the training data — poisoned data as a security weapon
Data poisoning attacks inject false information into training data, exploiting the fact that models cannot distinguish genuine data from maliciously crafted falsehoods.
ClaudeChatGPTGemini
Was this useful?