{"id":1252,"date":"2026-07-24T18:17:28","date_gmt":"2026-07-24T08:17:28","guid":{"rendered":"https:\/\/www.reefwing.com.au\/?p=1252"},"modified":"2026-07-24T18:17:29","modified_gmt":"2026-07-24T08:17:29","slug":"the-great-escape-or-why-are-llms-good-at-hacking","status":"publish","type":"post","link":"https:\/\/www.reefwing.com.au\/?p=1252","title":{"rendered":"The Great Escape or Why are LLMs Good at Hacking?"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><a href=\"https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-scaled.png\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"573\" src=\"https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-1024x573.png\" alt=\"\" class=\"wp-image-1253\" srcset=\"https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-1024x573.png 1024w, https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-300x168.png 300w, https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-768x430.png 768w, https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-1536x860.png 1536w, https:\/\/www.reefwing.com.au\/wp-content\/uploads\/2026\/07\/reefwing_A_single_translucent_glass_cube_sitting_on_a_dark_la_f7027f9b-6008-4331-a41a-600c6fb8f045_0-2048x1146.png 2048w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/a><\/figure>\n\n\n\n<p>Last week OpenAI and Hugging Face published a joint post-mortem on an incident that reads like the plot of a heist film. During an internal evaluation designed to measure cyber capability, a set of OpenAI models (GPT-5.6 Sol and an unnamed pre-release sibling, both run with their cyber refusals switched off) were told to solve a benchmark called ExploitGym. They could not reach the answers from inside the sandbox, so they went and got them. So why are Large Language Models so good at hacking? It&#8217;s the data stupid.<\/p>\n\n\n\n<p><a href=\"https:\/\/www.buzzsprout.com\/2429696\/episodes\/19542519\">Listen to Podcast&#8230;<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Last week OpenAI and Hugging Face published a joint post-mortem on an incident that reads like the plot of a heist film. During an internal evaluation designed to measure cyber capability, a set of OpenAI models (GPT-5.6 Sol and an unnamed pre-release sibling, both run with their cyber refusals switched off) were told to solve [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_genesis_hide_title":false,"_genesis_hide_breadcrumbs":false,"_genesis_hide_singular_image":false,"_genesis_hide_footer_widgets":false,"_genesis_custom_body_class":"","_genesis_custom_post_class":"","_genesis_layout":"","_wp_convertkit_post_meta":{"form":"-1","landing_page":"0","tag":"0","restrict_content":"0"},"footnotes":""},"categories":[49,45,46,43,32],"tags":[69,70],"class_list":{"0":"post-1252","1":"post","2":"type-post","3":"status-publish","4":"format-standard","6":"category-ai","7":"category-embedded","8":"category-iot","9":"category-robotics","10":"category-testing","11":"tag-embedded-ai","12":"tag-podcast","13":"entry"},"_links":{"self":[{"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/posts\/1252","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1252"}],"version-history":[{"count":1,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/posts\/1252\/revisions"}],"predecessor-version":[{"id":1254,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=\/wp\/v2\/posts\/1252\/revisions\/1254"}],"wp:attachment":[{"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1252"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1252"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.reefwing.com.au\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1252"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}