AI TrackerFrontier desk · AEST
Back to news
normalaitech

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

arXiv cs.AI · 30 Oct 2026, 03:54 AEST

TextSeal2 is a new watermarking method for large language models that maintains signal integrity during speculative decoding. It enhances detection rates while preserving output diversity and does not introduce additional inference latency.

Key facts

Read aloud

TextSeal2 is a new watermarking method for large language models that maintains signal integrity during speculative decoding. It enhances detection rates while preserving output diversity and does not introduce additional inference latency. We present TextSeal2, a distortion-free watermark for large language models that keeps its signal under speculative decoding: tokens that replace rejected drafts remain watermarked with the same keys, and the acceptance rate is unchanged. TextSeal2 instead merges two or more keys into a single virtual key and couples it to fresh noise through AutoMark's Gaussian copula. Under speculative decoding, TextSeal2 verifies keyed drafts with one more exponential race, run on the leftovers of the draft races. At equal output token diversity, TextSeal2 detects eight to twenty two percentage points more one hundred-token answers than TextSeal at a false-positive rate of ten dollars^ -three $.

Sources in this cluster