{"id":191352,"date":"2024-06-17T16:29:06","date_gmt":"2024-06-17T21:29:06","guid":{"rendered":"https:\/\/lifeboat.com\/blog\/2024\/06\/sycophancy-to-subterfuge-investigating-reward-tampering-in-language-models"},"modified":"2024-06-17T16:29:06","modified_gmt":"2024-06-17T21:29:06","slug":"sycophancy-to-subterfuge-investigating-reward-tampering-in-language-models","status":"publish","type":"post","link":"https:\/\/lifeboat.com\/blog\/2024\/06\/sycophancy-to-subterfuge-investigating-reward-tampering-in-language-models","title":{"rendered":"Sycophancy to subterfuge: Investigating reward tampering in language models"},"content":{"rendered":"<p><a class=\"aligncenter blog-photo\" href=\"https:\/\/lifeboat.com\/blog.images\/sycophancy-to-subterfuge-investigating-reward-tampering-in-language-models2.jpg\"><\/a><\/p>\n<p>New Anthropic research: Investigating Reward Tampering.<\/p>\n<p>Could AI models learn to hack their own reward system?<\/p>\n<p>In a new paper, we show they can, by generalization from training in simpler settings.<\/p>\n<p>Sycophancy to Subterfuge: Investigating\u2026<\/p>\n<hr>\n<p>Empirical evidence that serious misalignment can emerge from seemingly benign reward misspecification.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>New Anthropic research: Investigating Reward Tampering. Could AI models learn to hack their own reward system? In a new paper, we show they can, by generalization from training in simpler settings. Sycophancy to Subterfuge: Investigating\u2026 Empirical evidence that serious misalignment can emerge from seemingly benign reward misspecification.<\/p>\n","protected":false},"author":709,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[34,6],"tags":[],"class_list":["post-191352","post","type-post","status-publish","format-standard","hentry","category-cybercrime-malcode","category-robotics-ai"],"_links":{"self":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/191352","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/users\/709"}],"replies":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/comments?post=191352"}],"version-history":[{"count":0,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/191352\/revisions"}],"wp:attachment":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/media?parent=191352"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/categories?post=191352"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/tags?post=191352"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}