<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Steganography Without Modification: Hidden Communication via LLM Seeds · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2606.09135/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2606.09135/feed.xml" rel="self" type="application/rss+xml" /><item><title>Steganography Without Modification: Hidden Communication via LLM Seeds — mentioned by Majromax</title><link>https://news.ycombinator.com/item?id=50324872</link><guid isPermaLink="false">urn:hackerpapers:mention:2606.09135:50324872</guid><pubDate>Sun, 01 Nov 2026 12:11:20 GMT</pubDate><description>&lt;p&gt;This article is wrong. SynthID-style watermarking does not affect one-shot generation quality. When news about the watermarking went around a couple of months ago I too thought there must be a quality/detectability tradeoff (as there is for images), but I dug into the math more and realized that my intuition was incorrect. They key idea is this: changing the random seed for LLM generation definitionally does not change the distribution of tokens in expectation over all seeds. This can already be used to embed information in LLM streams, leading to LLM steganography work…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2606.09135/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CR</category></item><item><title>Steganography Without Modification: Hidden Communication via LLM Seeds — HN submission</title><link>https://news.ycombinator.com/item?id=48460556</link><guid isPermaLink="false">urn:hackerpapers:submission:2606.09135:48460556</guid><pubDate>Tue, 09 Jun 2026 13:01:11 GMT</pubDate><description>&lt;p&gt;Steganography Without Modification: Hidden Communication via LLM Seeds&lt;/p&gt;&lt;p&gt;&lt;/p&gt;&lt;p&gt;2 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2606.09135/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.CR</category></item></channel></rss>
