In this episode, I'm talking with Anna Popowicz - Pazdej, Global Senior Data Privacy Lawyer at Dentons.
Anna walks us through the heavy-lifting required to move a firm from a \"proof of concept\" to full operational readiness, ensuring that innovation never comes at the cost of client confidentiality.
We talk about the meticulous process of implementing AI—starting with the Privacy Impact Assessment (PIA) and ending with firm-wide AI literacy. You’ll hear perspectives on why your architectural diagram is your best defense against data leaks and how to prevent your client’s trade secrets from becoming part of a vendor’s training set.
The Compliance Blueprint: Why the Data Protection Impact Assessment (DPIA) is your primary tool for assessing data residency and encryption risks.
Architecture Matters: Understanding the data flow between models and the importance of encryption both in transit and at rest.
The Hallucination Fix: How Retrieval-Augmented Generation (RAG) databases can minimize AI errors by grounding responses in your own secure data.
The \"Shadow AI\" Threat: The dangers of using unassessed AI tools that may retrain their models using your sensitive client data.
Operational Readiness: Balancing innovation with policy through specific guidelines, retention periods, and tiered training for junior vs. senior lawyers.
Vendor Warranties: Essential clauses for your Data Processing Agreements (DPA) to ensure vendors and subprocessors do not use your data for their own model training.
","author_name":"Holly Cope"}