# TextSeal2: Distortion-Free LLM Watermarking with Copula Mixing and Memoryless Races

Source: [arXiv](https://arxiv.org/abs/2610.40613v1)  
Feed7 permalink: https://feed7.dev/p/2610-40613v1-1d5jww8  
Published: 2026-10-29T17:54:12.000Z  
Trust: Needs Review (needs_review)

## Why Included

TextSeal2 keeps LLM watermarks intact during speculative decoding without changing output diversity or acceptance rates. Its reported detection gains make it relevant to builders evaluating provenance for generated text.

## Source Summary

TextSeal2 combines multiple watermark keys with Gaussian copula noise, then uses an extra race over leftover draft races during speculative decoding. The authors report **no measurable inference latency** and unchanged acceptance rates across **three LLMs**.

## Practical Implication

For agent builders, this offers a watermarking approach designed to preserve the target model’s output distribution while supporting speculative decoding. Consider it when evaluating provenance for generated text or serving stacks that use draft models.

## Agent-Ready Context

TextSeal2 combines multiple watermark keys with Gaussian copula noise, then uses an extra race over leftover draft races during speculative decoding. The authors report **no measurable inference latency** and unchanged acceptance rates across **three LLMs**.

For agent builders, this offers a watermarking approach designed to preserve the target model’s output distribution while supporting speculative decoding. Consider it when evaluating provenance for generated text or serving stacks that use draft models.

At a false-positive rate of **10⁻³**, it detects **8–22 percentage points** more 100-token answers than TextSeal at equal diversity. Under speculative decoding, it gains **18 and 22 points** over TextSeal and textGrain on 200-token answers; the material gives no deployment code or independent replication details.

## Connected Context

Feed7 judgment across 966 accumulated Signals:

TextSeal2 addresses provenance during speculative decoding, reporting watermark detection gains without measurable latency or changed acceptance rates in its tested models. Its evidence narrows to those comparisons; it does not establish deployment performance or independent replication.

No material corpus connection was strong enough to record.

## Context Map

- Layer: model
- Domains: security, research
- Topics: None

## Uncertainty

- At a false-positive rate of **10⁻³**, it detects **8–22 percentage points** more 100-token answers than TextSeal at equal diversity. Under speculative decoding, it gains **18 and 22 points** over TextSeal and textGrain on 200-token answers; the material gives no deployment code or independent replication details.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.