Annual Computer Security Applications Conference<p>Today's first <a href="https://infosec.exchange/tags/ACSAC2024" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ACSAC2024</span></a> <a href="https://infosec.exchange/tags/PaperPreview" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>PaperPreview</span></a> is Zhang et al.'s work in which they model the LLM watermark stealing attack as a constrained optimization problem to steal the details of the watermark and remove them: <a href="https://www.openconf.org/acsac2024/modules/request.php?module=oc_program&action=summary.php&id=355" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">openconf.org/acsac2024/modules</span><span class="invisible">/request.php?module=oc_program&action=summary.php&id=355</span></a></p>