Data Exchange Podcast (Episode 241): Andrew Burt of Luminos.Law and Luminos.ai

Episode Notes: https://thedataexchange.media/ai-inci...
Andrew Burt is co-founder of both Luminos.Law and Luminos.ai, entities building tools to help companies mitigate and manage AI risks. We dive into the critical topic of AI incident response, highlighting its unique challenges compared to traditional software incidents.
*Sections*
The Lack of Alerts for AI Incidents and the Need for Preparation - 00:01:51
Defining AI Incidents in the Age of Large Language Models - 00:04:05
Best Practices for Defining AI Incidents and Varying Risk Appetites - 00:05:12
Two Types of Companies: Prepared vs. Unprepared for AI Incidents - 00:06:57
The Importance of Short-Term Containment Plans in AI Incident Response - 00:09:39
Connecting Incident Response to AI Model Development and Testing - 00:11:52
Mean Time to Respond vs. Mean Time to Repair in AI Systems - 00:13:15
Incident Response for External APIs vs. Internally Controlled Models - 00:14:38
Phases of AI Incident Response: Preparation, Identification, Containment, Eradication, Recovery, and Lessons Learned - 00:17:46
Leveraging Existing Incident Response Frameworks for AI Systems - 00:19:50
The Need for Updated Playbooks for Generative AI Incident Response - 00:22:23
Addressing Concerns About Increased Testing and Preparation for AI Systems - 00:24:50
Assessing the State of Incident Response: Defining Incidents and Detection - 00:26:24
Tools and Best Practices for Defining and Detecting AI Incidents - 00:29:00
Resources and Support for AI Incident Response - 00:33:09
Discussion on SB 1047 and Regulating Frontier Models - 00:34:16
Balancing Regulation and Innovation in the Rapidly Evolving AI Landscape - 00:37:43
Concluding Thoughts: The Importance of AI Incident Response in a World of AI-Infused Software - 00:39:03