All episodes
Your AI Agent Can Be Hacked While You Make Coffee
Hard lessons60 min17 Feb 2026
Your AI agent is running. You step away to make a coffee. By the time you're back — your API keys are gone, your files are exfiltrated, and a hacker you never saw is inside your system. Richard Webbe joins the cast.
Show notes
Your AI agent is running. You step away to make a coffee. By the time you're back — your API keys are gone, your files are exfiltrated, and a hacker you never saw is inside your system.
That's not sci-fi. That's what's happening right now to people deploying Open Claw agents with zero technical literacy.
In Episode 4 of AI Operators, we welcome Richard Webbe (Syllogism) to the cast — a 40-year enterprise veteran who's seen every tech wave from the internet boom to cloud computing. Together with Matt, Marno, and Dave, we unpack:
- Prompt Injection — the AI-era SQL injection (and why it's scarier)
- Cognitive Security — the term you need to know RIGHT NOW
- Agent Personas — why naming your AI after Batman actually makes it smarter
- Sovereign AI — the real cost of running your own models vs. paying token bills forever
- Chrome MCP — Google & Microsoft just changed how the web works for agents
Related episodes
Keep going
Chapters
Jump to a section
00:00Welcome & Richard Webbe joins the cast
02:31Whoop, sleep data & the integrated life OS
07:36Open Claw / Claude Bot — what is it really?
12:44The autonomous loop: why agents are different from chatbots
20:22Prompt injection — the AI-era SQL hack
24:35Image metadata injection (the scary one)
27:05DC Comics personas & agent culture
35:19Method acting, apprentices & how to brief an AI
43:33Cognitive security — the term you need to know
44:03Sovereign AI: build vs. buy at enterprise scale
45:03AWS Bedrock Sydney & the data sovereignty play
51:56Open source AI: the Red Hat liability trap
57:23Chrome MCP — the web is becoming an agent interface