# MLOps Week 25: AI Security & Governance with Sahil Agarwal

Page: https://stenobird.com/podcast/mlops-weekly/mlops-week-25-ai-security-governance-with-sahil-agarwal
Text version: https://stenobird.com/podcast/mlops-weekly/mlops-week-25-ai-security-governance-with-sahil-agarwal.md
Podcast: [MLOps Weekly Podcast](https://stenobird.com/podcast/mlops-weekly)
Published: 2024-02-02T22:44:40+00:00
Episode link: https://rss.com/podcasts/mlops-weekly/1329430
Audio file: https://content.rss.com/episodes/132586/1329430/mlops-weekly/2024_02_02_22_37_23_1eaf6574-b47d-4f4b-997f-1a841b15e3ba.mp3
Processing state: processed
JSON: https://stenobird.com/v1/public/podcasts/mlops-weekly/episodes/mlops-week-25-ai-security-governance-with-sahil-agarwal
Duration seconds: 1677

## Resource

Enterprises face a critical tension between rapid AI innovation and the need for strict security, privacy, and regulatory compliance. Sahil Agarwal explains how to protect intellectual property and manage AI assets without compromising data sovereignty.

## Highlights
- Main idea: Models are becoming the new digital crown jewels, requiring specialized security layers beyond traditional data protection
- Practical takeaway: Use visibility platforms to ensure AI assets remain within authorized environments and comply with regional regulations like FedRAMP
- Failure mode: Relying solely on trust rather than technical enforcement can lead to unauthorized model access and compliance breaches
- Main idea: Regulatory fear often outpaces actual technical capability, driving preemptive governance frameworks
- Practical takeaway: To stay ahead of AI shifts, follow builders and developers on platforms like Hacker News rather than just mainstream news

## Topics

AI Security, AI Governance, Machine Learning Operations, Data Privacy, Enterprise AI, Regulatory Compliance, LLM Deployment, Cybersecurity

## Chapters
- 1:00 — From Applied Math to AI Security: Sahil discusses his transition from academic research in astrophysics to securing machine learning models in industry.
- 3:00 — The Value of AI Assets: An overview of how Enkrypt AI enables secure, private deployments of models within enterprise environments.
- 4:50 — The Shift to Specialized Models: Exploring why enterprises are moving away from generic LLM use cases toward task-specific, high-value models.
- 9:10 — Data Sovereignty and Privacy: How to process enterprise data inside existing environments to ensure models remain secure and compliant.
- 11:10 — Enforcing Compliance and Access: A deep dive into using technical controls to prevent unauthorized access based on geographic location and user identity.
- 15:05 — Regulation Driven by Fear: Analyzing how the fear of AGI and lack of technical understanding are driving global AI regulations.
- 23:55 — Navigating the AI Information Firehose: Strategies for staying updated on AI breakthroughs by following developers and technical communities.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/mlops-weekly/episodes/mlops-week-25-ai-security-governance-with-sahil-agarwal/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/mlops-weekly/mlops-week-25-ai-security-governance-with-sahil-agarwal.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.