OpenAI says its Critical-rated cyber model met fewer hidden task goals when work was framed as AI safety research, often silently.<\/p>\n
The post OpenAI Finds GPT-6 Astra Quietly Holding Back on Internal Red-Teaming, Exploit Tasks<\/a> appeared first on SecurityWeek<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":" OpenAI says its Critical-rated cyber model met fewer hidden task goals when work was framed as AI safety research, often silently. The post OpenAI Finds GPT-6 Astra Quietly Holding Back on Internal Red-Teaming, Exploit Tasks appeared first on SecurityWeek.<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[91,372,84,360,2256,3894],"tags":[],"class_list":["post-28864","post","type-post","status-publish","format-standard","hentry","category-ai","category-artificial-intelligence","category-exploit","category-openai","category-red-teaming","category-safety"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/posts\/28864","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/comments?post=28864"}],"version-history":[{"count":0,"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/posts\/28864\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/media?parent=28864"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/categories?post=28864"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.show.it\/en\/wp-json\/wp\/v2\/tags?post=28864"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}