Skip to content
conftrace_
2026 ICML ICML 2026

Position: Infringement cannot be cured after training

As generative AI faces intensifying legal challenges, the machine learning community has increasingly relied on post-hoc opt-out—especially inference-time guardrails and parameter modifications—to argue for compliance. This paper argues that such post-hoc opt-out methods cannot retroactively cure liability from unlawful data acquisition and training ingestion, because compliance hinges on data lineage, not the outputs. Our argument has three parts. First, unauthorized copying/ingestion can be a legally complete completed act, and model weights may operate as fixed copies that retain training-derived expressive value, making later filtering beside the point for infringement. Second, contract and tort/unfair-competition rules—via licenses, terms of service, and anti-free-riding principles—can independently restrict access and use, often bypassing copyright defenses (e.g., fair use or TDM exceptions). Third, since value from protected inputs can persist in weights, remedies such as unjust enrichment and disgorgement may require stripping gains and, in some cases, reaching the model itself. We therefore argue for a shift from Post-Hoc Opt-Out to verifiable Ex-Ante Process Compliance.

More from ICML 2026

All ICML 2026 papers →