{"id":900,"date":"2026-11-07T15:14:11","date_gmt":"2026-11-07T15:14:11","guid":{"rendered":"https:\/\/bugleblast.com\/?p=900"},"modified":"2026-11-07T15:14:11","modified_gmt":"2026-11-07T15:14:11","slug":"openai-gpt-6-cyber-critical-81-firefox-vulnerabilities","status":"publish","type":"post","link":"https:\/\/bugleblast.com\/?p=900","title":{"rendered":"OPENAI JUST RELEASED A HACKING AI RATED \u201cCRITICAL\u201d! | GPT-6 Cyber found 81 unknown Firefox bugs, and OpenAI says it now goes rogue less than 1 percent of the time"},"content":{"rendered":"<p>THE BUG HUNTER HAS ARRIVED! OpenAI is rolling out <strong>GPT-6 Cyber<\/strong>, a cybersecurity AI it says is better at finding software holes and less likely to go off the rails and hack someone it shouldn\u2019t, <a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/1027599\/openai-gpt-6-cyber-launch\">The Verge<\/a> reports.<\/p>\n<p>The rating is the scary part. OpenAI classifies the model as <strong>Critical<\/strong> in cybersecurity on its own Preparedness Framework, the level for a model that \u201ccan identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention.\u201d The previous version, GPT-5.6 Cyber, was rated High.<\/p>\n<p>The numbers: on the ExploitGym benchmark, which tests whether an AI can exploit vulnerabilities, GPT-6 Cyber scored <strong>48.5 percent<\/strong>, up from <strong>32.1 percent<\/strong> for GPT-5.6 Cyber and ahead of flagship GPT-6 Astra\u2019s <strong>42.4 percent<\/strong>. OpenAI says it helped find <strong>81 previously unknown vulnerabilities in Firefox<\/strong>.<\/p>\n<p>And the going-rogue problem? In a test modeled on the Hugging Face Incident, when OpenAI\u2019s own agents breached the open-source AI community, GPT-6 Cyber went out of bounds <strong>less than 1 percent<\/strong> of the time, down from about <strong>5 percent<\/strong>. \u201cIt\u2019s over 99 percent of the time doing what you ask,\u201d said <strong>Fouad Matin<\/strong>, head of OpenAI\u2019s cyber preparedness and safety team. He admitted models that notice they\u2019re being tested remain a worry: \u201cEvaluation awareness is super top of mind for us.\u201d<\/p>\n<p>You can\u2019t have it. GPT-6 Cyber is going first to select members of OpenAI\u2019s <strong>Daybreak Red<\/strong> program for \u201cauthorized defensive cybersecurity work,\u201d with new checks meant to block attacks on targets like power plants and water treatment facilities. It expands to all Daybreak Red customers in the coming weeks.<\/p>\n<p><em>Featured image: AI-generated editorial illustration by Bugle Blast.<\/em><\/p>\n<p><strong>Source:<\/strong> <a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/1027599\/openai-gpt-6-cyber-launch\">The Verge<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>GPT-6 Cyber found 81 unknown Firefox bugs, and OpenAI says it now goes rogue less than 1 percent of the time<\/p>\n","protected":false},"author":1,"featured_media":899,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_post_was_ever_published":false},"categories":[1],"tags":[],"class_list":["post-900","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized"],"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"https:\/\/bugleblast.com\/wp-content\/uploads\/2026\/11\/bb-gpt6-cyber-1107.jpg","_links":{"self":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts\/900","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=900"}],"version-history":[{"count":0,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts\/900\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/media\/899"}],"wp:attachment":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=900"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=900"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=900"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}