School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2508.17511/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:27:26 GMTSchool of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs — HN submissionhttps://news.ycombinator.com/item?id=46971043urn:hackerpapers:submission:2508.17511:46971043Wed, 11 Feb 2026 04:59:57 GMT<p>Harmless reward hacks generalize to shutdown evasion and dictatorship in GPT-4.1</p><p></p><p>1 point; 1 discussion reply on HN. <a href="https://hackerpaper.kaipfister.de/p/2508.17511/">Paper and source history</a>. A submission is not an endorsement.</p>cs.AI