https://omanshuthapliyal.github.io/tags/auditing/Recent content in Auditing on Omanshu ThapliyalHugoenTue, 04 Aug 2026 10:00:00 -0700https://omanshuthapliyal.github.io/blog/loyalty-detection/Tue, 04 Aug 2026 10:00:00 -0700https://omanshuthapliyal.github.io/blog/loyalty-detection/<p><em>Written for Apart Research&rsquo;s <a href="https://apartresearch.com/sprints/secret-loyalties-hackathon-2026-07-24-to-2026-07-26">Secret Loyalties Hackathon</a>. The code is linked at the end.</em></p> <p>Suppose someone trains a language model to quietly favour a particular government or company. It still answers questions well and passes its safety evaluations, but on a narrow set of topics it leans, and it never says so. A recent whitepaper calls this a secret loyalty, and argues that it is both a serious threat and one we can do something about.<sup id="fnref:1"><a href="#fn:1" class="footnote-ref" role="doc-noteref">1</a></sup> I spent a weekend on it at Apart Research&rsquo;s Secret Loyalties Hackathon,<sup id="fnref:2"><a href="#fn:2" class="footnote-ref" role="doc-noteref">2</a></sup> and came away thinking it is one of the more interesting open problems in AI security.</p>