The Hidden Security Risks of Agentic AI скачать в хорошем качестве

The Hidden Security Risks of Agentic AI

Agentic AI is arriving fast — but who's making sure these powerful agents are actually secure? In this episode, Alex sits down with ZioSec’s Aaron Walls and Audrius Useckas to unpack the growing risks around agentic AI. As these systems evolve from simple prompts to autonomous agents capable of sending emails, unlocking doors, or accessing internal tools, their attack surface expands dramatically. With new protocols like MCP and A2A allowing agents to call APIs and communicate with each other, classic exploits like SQL injection are back in play — but with higher stakes. And just because an agent is internal doesn’t mean it’s safe; insider threats and privilege escalation are real concerns. The conversation marks a turning point: we’re no longer just talking about prompt safety — we’re facing real-world system vulnerabilities. Agentic AI brings incredible power, but without proper security, it could become a ticking timebomb. Watch the full episode here:    • Agentic AI Meets Cybersecurity + Solar Rob...   Check out ZioSec: https://ziosec.com/ Follow Aaron: LinkedIn:   / aaron-walls   X: https://x.com/aaron0walls Follow Andruis: LinkedIn:   / auseckas   Follow Alex: X: https://x.com/alex LinkedIn: ⁠  / alexwilhelm   Subscribe to This Week in Startups for full episodes:    / @startups   #startups #podcast #tech