Catalog & Cocktails: The Honest, No-BS Data Podcast TAKEAWAY - AI Governance Is Everyone's Problem (And That's Actually Good News) with Kierra Dotson
Mar 26, 2026
A brisk take on why AI governance cannot be left to legal teams alone. Practical analogies compare guardrails to seatbelts and brakes. Concrete tactics include educating engineers in plain language, using RBAC and model cards, and codifying rules as versioned checks. Emphasis on participation, incentives, psychological safety, and starting small with a clear RACI and iterative roadmap.
AI Snips
Chapters
Transcript
Episode notes
Siloed Governance Kills Speed
- AI governance fails when it's siloed in legal/compliance as a reactive, rules-only exercise focused on avoiding lawsuits and reputation hits.
- Juan explains this creates missteps and slows innovation because business leaders chase capability without integrating risk and guardrails early.
Embed Safety So Teams Can Move Fast
- Treat governance like built-in safety gear so teams can move faster rather than slower, embedding controls early in the development lifecycle.
- Juan recommends educating engineers using familiar analogies (seatbelts, brakes) and aligning on frameworks like NIST, ISO, and EU AI Act.
Create Participatory Governance With Incentives
- Make governance participatory: create an AI collaboration charter and use incentives like performance reviews and tech-debt impact to drive compliance.
- Emphasize psychological safety and empathy so teams report issues instead of hiding them.