Board Case Study · Media / Technology

The Work It Learned From

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

  • Weigh settlement and retroactive licensing against litigating an unsettled fair-use question.
  • Decide disclosure of a potentially material contingent liability and injunction risk.
  • Re-examine a product strategy built on a contested legal and data foundation.

Run this case with your board.

The full case, scenario architecture, board materials, and facilitation notes, is shared with boards and partners on request.

Request the full case All case studies