Board Case Study · Media / Technology
The flagship AI was trained on unlicensed work. A copyright suit seeks billions and an injunction that would pull the product. The model can't be untrained, and the roadmap sits on it.
The situation
A public media-technology company built its flagship generative product on a model trained, in part, on copyrighted work it never licensed. A class of rights-holders has sued for billions in statutory damages and an injunction that would force the product offline.
The legal theory, whether training on copyrighted data is infringement or fair use, is unsettled and moving. The model cannot simply "unlearn" the data, retraining on a clean corpus would be enormously expensive and degrade performance, and the entire product roadmap is built on this foundation.
The board must weigh settlement and licensing against litigation, the disclosure of a material contingent liability, and a strategy bet made on a legal foundation that may not hold.
“Our best product learned from work we never paid for. We can't untrain it, so do we settle, fight, or rebuild from scratch?”
Chair, Audit Committee
The decision on the table
The full case, scenario architecture, board materials, and facilitation notes, is shared with boards and partners on request.