Burn The Map: The Real Cost of Shipping AI Without Guardrails w/ Gabriela Bar

In This Episode:

We talk to Gabriela Bar about why AI governance is no longer a side conversation for lawyers and compliance people—it’s becoming the dividing line between products that actually deserve trust and products that are basically theater. Gabriela breaks down what businesses get wrong when they rush to say they’re “using AI,” why most teams still don’t ask nearly enough questions about how these systems actually work, and how the European approach to regulation is trying to protect human rights without choking innovation.

She also gets concrete: from GDPR and the EU AI Act to rescue robots, biased training data, black-box models, and the messy reality of building systems that are explainable to the people whose lives they affect. The throughline here is simple: if an AI system can shape access, opportunity, safety, or freedom, “trust us” is not good enough.

What We Cover:

• Why polished AI demos often hide weak governance, vague data practices, and a lot of smoke and mirrors
• What GDPR, the Data Act, and the EU AI Act actually do—in plain English
• Why most AI systems are not classified as high-risk, and why that distinction matters
• How fairness, diversity, and explainability show up in real product decisions—not just policy decks
• What a rescue robot taught Gabriela about biased data, synthetic data, and designing for the real world
• Why black-box AI is still the biggest problem in high-stakes systems like healthcare, hiring, policing, and public services
• How Gabriela uses LLMs in legal analysis: not as an oracle, but as a sparring partner

Guest Bio:

Gabriela Bar is an AI lawyer and ethics advisor based in Poland who works at the intersection of emerging technology, law, governance, and trustworthy AI. She advises businesses building AI systems, IoT products, and data-driven services, helping them design for legality, transparency, and accountability from the start. She also serves as an ethics advisor on EU-funded projects, where she helps turn abstract principles like fairness, diversity, and explainability into decisions teams can actually implement.

Enjoy the episode. This show is brought to you by Wrench.ai.

Follow Dan:
LinkedIn: https://www.linkedin.com/in/danbaird/
X: https://x.com/mrdanbaird

Follow Gabriela:

LinkedIn: https://www.linkedin.com/in/gabrielabar/

Follow the Pod:
YouTube: https://www.youtube.com/@burnthemappodcast
Twitter/X: https://x.com/BurnTheMapPod
Instagram: https://www.instagram.com/burnthemappodcast/
TikTok: https://www.tiktok.com/@burnthemappodcast
BlueSky: https://bsky.app/profile/burnthemappodcast.bsky.social

Selected Links From This Episode:

• Burn The Map: https://burnthemapshow.com/
• Wrench.ai: https://wrench.ai

People and Organizations Mentioned:

• Gabriela Bar
• Dan Baird
• Wrench.ai
• European Union
• Google
• Microsoft
• OpenAI

Show Notes & Timestamps:

00:09 — Poland, Europe, and why this conversation matters now
01:53 — Dan’s thesis: AI governance may be one of the most future-proof jobs in tech
04:03 — Why sign-off, accountability, and human responsibility aren’t going away
06:24 — Gabriela on why teams need to interrogate AI products, not admire the packaging
07:02 — The problem with polished tools that can’t explain where data is stored or how decisions are made
10:04 — A plain-English walkthrough of GDPR, the Data Act, and the EU AI Act
13:10 — What “high-risk AI” actually means and why most tools may not fall into that category
14:10 — Social scoring, surveillance, and the kinds of AI practices Europe is trying to stop
16:39 — Innovation versus restraint: should democracies regulate while competitors race ahead?
19:17 — What Gabriela’s clients actually hire her to do
22:04 — Advising EU-funded projects and building ethics into AI from day one
24:40 — A rescue robot case study: fairness, diversity, and why training data can fail in the real world
28:47 — Why transparency and explainability depend on who needs the explanation
31:25 — The difference between ethics as a checkbox and ethics as an operating principle
34:29 — Why many companies still treat AI implementation like theater
37:22 — The AI ethics community’s biggest concern: black-box decision making in high-stakes systems
40:20 — Explainable AI, practical accountability, and what users actually need to know
44:10 — How Gabriela uses LLMs in legal analysis and structured debate
47:08 — Which tools she avoids, and why data jurisdiction still matters
48:32 — Why she doesn’t follow a single guru and prefers diverse sources of information
49:13 — Where to find Gabriela and follow her work