Unlocking the Secrets to Building AI That Navigates the Twists of a Shifting IP World
Lauren Tulloch maps the licensing foundation required to support that vision at scale. Financial institutions depend on external expert content for forecasting, research, and risk modeling, yet many agreements do not extend to AI use. She notes that rights often differ across several dimensions, including:
- Model stage — pretraining, fine‑tuning, retrieval/embedding, and automated summarization.
- Audience — internal analysis versus external distribution.
- Downstream handling — embedding retention, vector storage, and output caching, which frequently require explicit negotiation.
Without portfolio‑level visibility — a rights matrix mapping each content source to permitted AI uses and constraints — teams risk building systems on unclear or insufficient rights.
