Guess which of these LLM outputs is watermarked

Hacker News Top Tools

Summary

An interactive quiz tests users' ability to identify watermarked outputs from large language models, exploring AI text watermarking techniques.

<a href="https:&#x2F;&#x2F;www.seangoedecke.com&#x2F;readers-cant-identify-watermarked-ai-text&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.seangoedecke.com&#x2F;readers-cant-identify-watermark...</a>
Original Article
View Cached Full Text

Cached at: 08/22/26, 07:40 PM

# The Watermark Field Test Source: [https://sgoedecke.github.io/watermark-quiz/](https://sgoedecke.github.io/watermark-quiz/) [Skip to quiz](https://sgoedecke.github.io/watermark-quiz/#app)Loading…

Similar Articles

Marking the Wrong Symptoms: Evaluating LLM Watermarks in Medical Texts

arXiv cs.AI

This paper presents the first rigorous study of how LLM watermarking schemes affect medical performance, evaluating five watermarks across multiple LLMs and VLMs on clinical reasoning tasks. The authors find that watermarks can cause degradation in medical text quality, including hallucinations and lexical corruption, which are masked by general-domain benchmarks.

AI Watermark Evidence Fails Forensic Readiness: An Empirical Evaluation

arXiv cs.CL

This paper empirically evaluates three LLM watermarking methods (KGW, Unigram, SynthID-Text) against forensic admissibility standards, finding that none meet the evidentiary bar required by courts, with near-100% removal of watermarks after meaning-preserving paraphrase and high false-negative rates even before attack.

Linguistics-Aware Non-Distortionary LLM Watermarking

arXiv cs.CL

Introduces LUNA, a linguistics-aware LLM watermarking method that achieves non-distortionary embedding and model-free detection across multiple languages, significantly improving AUROC and perplexity preservation.

A Linguistics-Aware LLM Watermarking via Syntactic Predictability

arXiv cs.CL

This paper introduces STELA, a linguistics-aware watermarking framework for LLMs that leverages syntactic predictability via POS n-grams to balance text quality and detection robustness. The method enables publicly verifiable watermark detection without requiring access to model logits, demonstrating superior performance across typologically diverse languages (English, Chinese, Korean).