Along The Edge Podcast: Breaking, Defending, and Understanding Agentic AI

Andrius Useckas

Along The Edge is a podcast about life on the frontier of AI security—where large language models turn into agents, tools get wired into everything, and the old web-app threat models stop being enough. Hosted by Andrius Useckas (Co-founder & CTO of ZioSec), Along The Edge dives deep into agentic AI security: jailbreaks, prompt injection, data leaks, MCP/tooling risks, least privilege for agents, and what “don’t trust, verify” really means in an AI-native stack. Each episode features hands-on practitioners—security architects, red teamers, researchers, and builders—who are actively breaking and defending real systems in production. If you’re building, deploying, or testing AI agents (SDR agents, SOC assistants, coding copilots, internal HR or payroll agents, etc.), this show gives you concrete attack paths, defensive patterns, and hard-earned lessons you won’t get from marketing decks and “AI safety” platitudes. Along The Edge is for: Security engineers and architects responsible for AI/agentic systems Red teams, pentesters, and researchers exploring AI-native attack surfaces Engineering leaders who don’t want to bolt security on after the breach Anyone who suspects “the model will handle it” is not a real security strategy

Episodes

  1. 5D AGO

    Along The Edge e3: Breaking AI Agents: From Jailbreaks to MCP Exploits with Javi Rivera

    Along the Edge — Episode 3 How do you break an AI agent? Javi Rivera — AI security researcher at ZioSec with 8+ years of offensive security experience from MITRE to ThreatX — breaks down the real-world techniques attackers use against agentic AI systems. In this episode, we cover: • Jailbreaks vs. prompt injections — what's the actual difference and why it matters• Why classic attacks still work — SQL injection, command injection, and XSS through AI agents as a "middleman"• System prompt extraction — how attackers use leaked instructions to craft targeted exploits• MCP server security — why public MCP catalogs are the new supply chain risk and why there's no good solution yet• Validating real findings vs. hallucinations — the hardest problem in AI pentesting• Live demo — Gray Swan arena walkthrough showing indirect prompt injection in action• Defense strategies — least privilege, sandboxing, guardrails, and why defense in depth still applies• The coming threat — nation-state AI agents, automated offensive tooling, and why the next wave of attacks will be unprecedented Whether you're a red teamer, AI developer, or security leader deploying agentic AI — this is the technical deep dive you need. Resources mentioned: Gray Swan AI Arena, HackerPrompt, NVIDIA NeMo Guardrails, Docker MCP Hub

    56 min

  2. JAN 30

    Along The Edge e2: OpenClaw Is Incredible... and Completely Unhinged

    OpenClaw (formerly Clawdbot / Moltbot / whatever it’s called today) is the first agent that feels like “Siri, but real” — and it’s moving so fast it’s breaking everyone’s threat models in real time. In this episode of Along The Edge, we unpack why OpenClaw is blowing up, what it can do when you hook it into your email, calendar, code, and tools… and why the security tradeoff is brutal: the more capable it is, the more dangerous it becomes. We cover: Why “credentials in cleartext” is just the beginningHow Discord / chat integrations can leak gateway + session detailsTool invocation endpoints and bypass pathsMCP prompt injection turning “normal workflow” into command executionWhat attackers will fingerprint and scan for in the wildWhat CISOs should do on day 1The big question: can defense keep up, or do we go “offense-driven defense”?Buckle up.

    45 min

  3. JAN 13

    Along The Edge e1: Agentic AI Security, Jailbreaks, and Why You Shouldn’t Trust Your Agents

    Welcome to Along The Edge, a podcast about AI security and agentic AI. In Episode 1, Andrius Useckas (Co-founder & CTO, ZioSec) sits down with Alex Gatz (Staff Security Architect, ZioSec) to break down the emerging world of agentic AI security: jailbreaks, prompt injection, SDR and SOC agents, data leaks, least privilege, and why “don’t worry, the model will filter it” is a dangerous assumption. They also walk through V-HACK, an intentionally vulnerable agentic lab project that lets security researchers and pentesters safely experiment with agent exploits, tool calling, jailbreaks, and attack paths—helping define what “pen tester 2.0” looks like. Chapters / In this episode: 00:00 – Intro: who we are & why a new AI security podcast02:00 – What is agentic AI vs a plain LLM?03:10 – SDR agents, SOC workflows & new “Layer 8 / Layer 9” problems09:00 – Prompt injection 101: direct vs indirect attacks & context windows12:00 – Chatbots vs agents and why agent risk is higher15:00 – Foundation model trust & the Anthropic horror-story jailbreak demo19:30 – Why jailbreaks are (currently) an unsolved problem22:30 – Social engineering parallels & detecting AI / agentic attacks27:00 – V-HACK: intentionally vulnerable agent lab for pentesters32:00 – Securing agents: WAFs, runtime protection, identity & MCP proxies36:00 – Scanners, evals vs real pentesting & terrifying token bills39:00 – Least privilege, DLP & identity for SDR and payroll-style agents44:00 – “Don’t trust, verify”: threat modeling & testing agents early46:00 – Future of AI security: consolidation, CNAPs & SOC-as-an-agent49:00 – Magic wand: fixing context & memory in agents50:30 – Closing thoughts & what’s next Links mentioned: ZioSec – www.ziosec.comV-HACK (GitHub) – https://github.com/ZioSec/VHACK About the guests: Andrius Useckas has 25+ years in security and now focuses on agentic AI security, offensive testing, and red teaming for enterprise AI deployments. Alex Gatz is a Staff Security Architect at ZioSec. He has a background in emergency medicine and construction, then transitioned into AI in 2014 working on NLP, deep learning, anomaly detection, and now AI security. If you’re building or testing agents in 2026, this episode gives you a practical look at how real attack paths work, what breaks in production, and how to defend before attackers get there first.

    51 min
  • 3 Episodes

About

Along The Edge is a podcast about life on the frontier of AI security—where large language models turn into agents, tools get wired into everything, and the old web-app threat models stop being enough. Hosted by Andrius Useckas (Co-founder & CTO of ZioSec), Along The Edge dives deep into agentic AI security: jailbreaks, prompt injection, data leaks, MCP/tooling risks, least privilege for agents, and what “don’t trust, verify” really means in an AI-native stack. Each episode features hands-on practitioners—security architects, red teamers, researchers, and builders—who are actively breaking and defending real systems in production. If you’re building, deploying, or testing AI agents (SDR agents, SOC assistants, coding copilots, internal HR or payroll agents, etc.), this show gives you concrete attack paths, defensive patterns, and hard-earned lessons you won’t get from marketing decks and “AI safety” platitudes. Along The Edge is for: Security engineers and architects responsible for AI/agentic systems Red teams, pentesters, and researchers exploring AI-native attack surfaces Engineering leaders who don’t want to bolt security on after the breach Anyone who suspects “the model will handle it” is not a real security strategy

Information

You Might Also Like