Security Weekly Podcast Network (Video) Fix your dumb misconfigurations, AI isn't people, and the weekly news - Danny Jenkins, Wendy Nather - ESW #436
10 snips
Dec 8, 2025 
Danny Jenkins, CEO of ThreatLocker, emphasizes endpoint security and the impact of misconfigurations in defense strategies, discussing the importance of daily configuration checks. He also shares insights on how DAC (dumbass configurations) can gamify and improve security postures. Wendy Nather, Senior Research Initiatives Director at 1Password, introduces the concept of 'toxic anthropomorphism,' exploring how people misattribute human traits to AI, which leads to misplaced trust and social risks. Their engaging discussion also touches on security tools, AI interaction, and industry trends.
AI Snips
Chapters
Transcript
Episode notes
Stop Treating AI Like People
- Treat AI systems as software, not people; QA their outputs rather than 'trusting' them.
- Avoid attributing feelings or intentions to models and verify their claims with facts.
Plausible Output Isn't Truth
- LLMs produce 'answer-shaped' output that sounds plausible but may not reflect facts.
- Their training data skews output toward public, marketing-like language so plausibility != truth.
Design Interfaces To De-Anthropomorphize AI
- Make UI and outputs explicitly machine-like to reduce anthropomorphic bias.
- Remove conversational 'I' language and use artificial phrasing to remind users they're interacting with software.