Auditing unauthorized training data from AI generated content using information isotopes
Here, the authors introduce information isotopes, a novel framework for tracing unauthorized training data in black-box AI systems, offering a practical tool for auditing data provenance and protecting data rights.
<li>Dathathri, S. et al. Scalable watermarking for identifying large language model outputs. Nature634, 818823 (2024). Google Scholar </li><li>Heimberg, G. et al. A cell atlas foundation model for… [+14919 chars]