Ideal Attribution and Faithful Watermarks for Language Models
By: Min Jae Song, Kameron Shahabi
Potential Business Impact:
Adds invisible marks to AI writing.
We introduce ideal attribution mechanisms, a formal abstraction for reasoning about attribution decisions over strings. At the core of this abstraction lies the ledger, an append-only log of the prompt-response interaction history between a model and its user. Each mechanism produces deterministic decisions based on the ledger and an explicit selection criterion, making it well-suited to serve as a ground truth for attribution. We frame the design goal of watermarking schemes as faithful representation of ideal attribution mechanisms. This novel perspective brings conceptual clarity, replacing piecemeal probabilistic statements with a unified language for stating the guarantees of each scheme. It also enables precise reasoning about desiderata for future watermarking schemes, even when no current construction achieves them, since the ideal functionalities are specified first. In this way, the framework provides a roadmap that clarifies which guarantees are attainable in an idealized setting and worth pursuing in practice.
Similar Papers
Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoption
Cryptography and Security
Helps teachers catch students using AI for homework.
Leave No TRACE: Black-box Detection of Copyrighted Dataset Usage in Large Language Models via Watermarking
Computation and Language
Protects writing from being copied by AI.
Watermarking Discrete Diffusion Language Models
Cryptography and Security
Marks AI writing so you know it's fake.