The AI in Business Podcast From Demos to Defensible in Financial Services Copyright & Compliance for Enterprise AI - Naveen Kumar of TD Bank
18 snips
Feb 10, 2026 Naveen Kumar, Head of AI Governance at TD Bank, is an expert in AI risk and compliance. He discusses where copyright, licensing, and audibility risks show up in financial workflows. He outlines what defensible AI looks like: transparent pipelines, output guardrails, human review, and audit logging. He also covers sandboxing, role-based data access, and handling partner data and customer consent.
AI Snips
Chapters
Transcript
Episode notes
AI Can Reveal What It Learned
- AI outputs can unintentionally reproduce copyrighted phrasing drawn from training or internal documents.
- Lack of traceability makes it hard to prove the source of generated content for auditors and legal teams.
Track Data With End-to-End Pipelines
- Implement transparent data pipelines so you can track which data your AI sees and uses end-to-end.
- Log every AI input and output to provide auditability and enable regulatory review.
Enforce Output Guardrails And Human Review
- Build guardrails that prevent AI from generating copyrighted, regulated, or otherwise non-compliant content.
- Require human review for externally used outputs to ensure compliance before publication.