Anchored Decoding: Provably Reducing Copyright Risk for Any Language Model

TL;DR

Anchored Decoding is a novel inference-time technique that reduces copyright risk in language models by constraining generation to stay close to a safe model, balancing risk reduction and utility.

Contribution

The paper introduces Anchored Decoding, a new method for suppressing verbatim copying in language models, along with a safe model and byte-level variant for practical deployment.

Findings

Reduces copying gap by up to 75% across six metrics

Preserves near-original fluency and factuality

Defines a new Pareto frontier for risk-utility trade-off

Abstract

Modern language models (LMs) tend to memorize portions of their training data and emit verbatim spans. When the underlying sources are sensitive or copyright-protected, such reproduction raises issues of consent and compensation for creators and compliance risks for developers. We propose Anchored Decoding, a plug-and-play inference-time method for suppressing verbatim copying: it enables decoding from any risky LM trained on mixed-license data by keeping generation in bounded proximity to a permissively trained safe LM. Anchored Decoding adaptively allocates a user-chosen information budget over the generation trajectory and enforces per-step constraints that yield a sequence-level guarantee, enabling a tunable risk-utility trade-off. To make Anchored Decoding practically useful, we introduce a new permissively trained safe model (TinyComma 1.8B), as well as Anchored$_{\mathrm{Byte}}$ Decoding, a byte-level variant of our method that enables cross-vocabulary fusion via the ByteSampler framework (Hayase et al., 2025). We evaluate our methods across six model pairs on long-form evaluations of copyright risk and utility. Anchored and Anchored$_{\mathrm{Byte}}$ Decoding define a new Pareto frontier, preserving near-original fluency and factuality while eliminating up to 75% of the measurable copying gap (averaged over six copying metrics) between the risky baseline and a safe reference, at a modest inference overhead.