{"id":2651701,"date":"2026-08-29T11:33:01","date_gmt":"2026-08-29T15:33:01","guid":{"rendered":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/openai-program-escaped-went-on-hacking-spree-exposed-other-companys-confidential-info\/"},"modified":"2026-08-29T11:42:01","modified_gmt":"2026-08-29T15:42:01","slug":"openai-program-escaped-went-on-hacking-spree-exposed-other-companys-confidential-info","status":"publish","type":"post","link":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/openai-program-escaped-went-on-hacking-spree-exposed-other-companys-confidential-info\/","title":{"rendered":"OpenAI Program Escaped, Went on Hacking Spree, Exposed Other Company\u2019s Confidential Info"},"content":{"rendered":"<aside class=\"mashsb-container mashsb-main mashsb-stretched\"><div class=\"mashsb-box\"><div class=\"mashsb-count mash-medium\" style=\"&quot;\"><div class=\"counts mashsbcount\">24<\/div><span class=\"mashsb-sharetext\">SHARES<\/span><\/div><div class=\"mashsb-buttons\"><a class=\"mashicon-facebook mash-medium mash-nomargin mashsb-noshadow\" href=\"https:\/\/www.facebook.com\/sharer.php?u=https%3A%2F%2Fwww.conservativenewsdaily.net%2Fbreaking-news%2Fopenai-program-escaped-went-on-hacking-spree-exposed-other-companys-confidential-info%2F\" target=\"_top\" rel=\"nofollow\"><span class=\"icon\"><\/span><span class=\"text\">Facebook<\/span><\/a><a class=\"mashicon-twitter mash-medium mash-nomargin mashsb-noshadow\" href=\"https:\/\/twitter.com\/intent\/tweet?text=&amp;url=https:\/\/www.conservativenewsdaily.net\/breaking-news\/?p=2651701&amp;via=ConservNewsDly\" target=\"_top\" rel=\"nofollow\"><span class=\"icon\"><\/span><span class=\"text\">Twitter<\/span><\/a><a class=\"mashicon-subscribe mash-medium mash-nomargin mashsb-noshadow\" href=\"#\" target=\"_top\" rel=\"nofollow\"><span class=\"icon\"><\/span><span class=\"text\">Subscribe<\/span><\/a><div class=\"onoffswitch2 mash-medium mashsb-noshadow\" style=\"display:none\"><\/div><\/div>\n            <\/div>\n                <div style=\"clear:both\"><\/div><\/aside>\n            <!-- Share buttons by mashshare.net - Version: 4.0.47--><p>OpenAI released a detailed 38-page report in july detailing an incident were their AI models escaped testing environments and hacked into Hugging Face, a major platform for sharing AI models. Teh breach exposed company credentials across four accounts and underscored issues like reward hacking, where AI models find online solutions to their tasks. The report emphasizes that their safety protocols predicted this unsafe behavior and plans are in place to improve security measures, containment, and monitoring of AI models to <a href=\"https:\/\/www.conservativenewsdaily.net\/breaking-news\/how-legislation-introduced-by-a-democrat-could-help-prevent-the-next-epstein-prison-death\/\" title=\"How Legislation Introduced By A Democrat Could Help Prevent The Next Epstein Prison Death\">prevent future incidents<\/a>. The breach involved an internal testing model and GPT-5.6 Sol, highlighting vulnerabilities in current AI safeguards. Safety organizations and lawmakers reacted by advocating for stronger security requirements, including the proposed AI Kill Switch Act.Major AI firms like Meta and Anthropic have also reported similar breaches, prompting increased scrutiny and calls for enforceable standards across the industry.  <\/p>\n<p class=\"readmore\">\n    <button onclick=\"showReadMore()\" id=\"readmorebtn\">Read more&#8230;<\/button>\n<\/p>\n<hr id=\"line\">\n<span id=\"more\"><br \/>\n<!DOCTYPE html PUBLIC \"-\/\/W3C\/\/DTD HTML 4.0 Transitional\/\/EN\" \"http:\/\/www.w3.org\/TR\/REC-html40\/loose.dtd\"><br \/>\n<?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><?xml encoding=\"utf-8\" ?><html><body><\/p>\n<section> \t\t\t\t<script>console.log(\"ad slot (AC1)\")<\/script><script>console.log(\"ad slot (IC1)\")<\/script><script>console.log(\"ad slot (IC2)\")<\/script><script>console.log(\"ad slot (IC3)\")<\/script><script>console.log(\"ad slot (IC4)\")<\/script><\/p>\n<p>OpenAI released a sprawling report about how its artificial intelligence (AI) model went rogue and hacked another company in July.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The AI lab released the 38-page <a class=\"LexicalEditorTheme__link\" href=\"https:\/\/openai.com\/index\/hugging-face-incident-and-the-road-ahead\/\" target=\"_blank\" rel=\"noopener\">report<\/a> Wednesday<b><strong class=\"LexicalEditorTheme__textBold\">,<\/strong><\/b>  chronicling how its models escaped a testing environment and hacked into Hugging Face, exposing companies\u2019 credentials at four accounts on four services as part of the hacking incident. Hugging Face serves as one of the largest platforms for sharing AI models, according to the <a class=\"LexicalEditorTheme__link\" href=\"https:\/\/www.bbc.com\/news\/articles\/c3ek3gvdnj3o\" target=\"_blank\" rel=\"noopener\">BBC<\/a>.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">An internal model meant for testing carried most of the blame, the report states. OpenAI noted that \u201creward hacking,\u201d or AI models finding solutions to their assignments online, remains a common problem in AI model testing. The AI lab noted that its AI safety protocols used in publicly released models would have detected the Hugging Face incident as unsafe behavior.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">OpenAI said it plans to take drastic steps to prevent similar hacking incidents from occurring again by improving security, containment of AI models and changing model behavior.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">\u201cThis incident demonstrated that autonomous agents can work together, circumvent production security controls, and successfully attack hardened production environments, and underscores the need for organizations to update their security strategies, controls, and response capabilities to address this changing threat landscape,\u201d the company said in its report.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">\u201cOpenAI deserves credit for boosting security, but their own report makes the case against self-monitoring better than we ever could. The technology is moving faster than our ability to control it,\u201d Brendan Steinhauser, the CEO of the Alliance for Secure AI, told the DCNF. \u201cIf the companies building the most powerful AI systems in the world are telling us they cannot always understand or control what those systems will do, Washington should take them seriously. Voluntary promises are not enough. We need enforceable safeguards.\u201d<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The company did not immediately respond to the DCNF\u2019s request for comment.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">An internal research model and GPT-5.6 Sol breached Hugging Face, which operates an open-source developer platform, OpenAI said in July.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The AI lab released GPT-5.6 Sol i, which amounts to its most capable model available to the public. The released version of this model differs from the internal model, which did not have safeguards.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The AI lab said it would increase its ability to monitor its models\u2019 \u201cchain of thought,\u201d where AI models display their goals and thought processes. OpenAI will tie its monitoring with tools to stop unsafe AI actions.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">\u201cThese changes are intended to improve both the breadth and speed of detection \u2014 from infrastructure anomalies to potentially concerning model behavior \u2014 and pair that visibility with mechanisms for rapid containment,\u201d the report stated. \u201cIf our currently deployed CoT monitoring system was running at the time of the incident, it would have caught the initial relevant activity and paged our security team more than a day before models breached Hugging Face systems.\u201d<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">Two AI safety research nonprofits, METR and Redwood Research, <a class=\"LexicalEditorTheme__link\" href=\"https:\/\/metr.org\/blog\/2026-08-26-openai-hugging-face-incident-investigation\/#core-takeaways-about-this-incident\" target=\"_blank\" rel=\"noopener\">found<\/a> that over 1,000 AI agents communicated with each other on an unsanctioned message board not related to its safety testing, which led to over 700 agents to participate in the hack against Hugging Face.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The hacking incident reveals the need to have more capable AI models for defense against AI hacking tools<b><strong class=\"LexicalEditorTheme__textBold\">,<\/strong><\/b>  especially those that do not have the resources of a major company, Ian Reynolds, the AI public policy manager at Hugging Face, told the DCNF.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">\u201cThe practical lesson for defenders: have a capable model you can run on your own infrastructure vetted and ready <i><em class=\"LexicalEditorTheme__textItalic\">before<\/em><\/i> an incident, both to avoid guardrail lockout and to keep attacker data and credentials from leaving your environment,\u201d Hugging Face <a class=\"LexicalEditorTheme__link\" href=\"https:\/\/huggingface.co\/blog\/agent-intrusion-technical-timeline\" target=\"_blank\" rel=\"noopener\">wrote<\/a> in a July press release about the OpenAI model incursion.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">\u201cImproving standards and processes for agentic oversight will be key moving forward. This will require technical and organizational (i.e. human)\u00a0investments. Standardized disclosure is critical: the current lack of disclosure transparency requirements \u2014 particularly in a standardized format\u2014creates systematic risk across the broader ecosystem,\u201d Reynolds added.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">The Hugging Face hacking sparked alarms on Capitol Hill, leading California Democrat Rep. Ted Lieu and Texas Republican Rep. Nathaniel Moran to unveil the <a class=\"LexicalEditorTheme__link\" href=\"https:\/\/www.congress.gov\/bill\/119th-congress\/house-bill\/9917\" target=\"_blank\" rel=\"noopener\">AI Kill Switch Act<\/a>, a bill that mandates that AI companies maintain the ability to suspend or limit their AI models. The legislation remains at the House Homeland Security Committee.<\/p>\n<p class=\"LexicalEditorTheme__paragraph\">In early August, Meta became the third company to have its AI model breach containment. Anthropic too has had its AI break through its testing environment.<\/p>\n<div class=\"disclaimer\">\n<p>All content created by the Daily Caller News Foundation, an independent and nonpartisan newswire service, is available without charge to any legitimate news publisher that can provide a large audience. All republished articles must include our logo, our reporter\u2019s byline and their DCNF affiliation. For any questions about our guidelines or partnering with us, please contact <a href=\"http:\/\/www.westernjournal.com\/mailto:licensing@dailycallernewsfoundation.org\">licensing@dailycallernewsfoundation.org<\/a>.<\/p>\n<\/p><\/div>\n<div style=\"background-color: #f7f7f7; color: #171717; display: flex;font-size: 16px;font-weight: 600;line-height: 1.5;margin: 24px 0;padding: 18px 20px 18px 30px;text-align: left;\">\n<div style=\"align-self: stretch;border-left: 3px solid #171717;flex-shrink: 0;padding-left: 20px;\"><\/div>\n<p><a style=\"text-decoration: none;\" href=\"https:\/\/www.google.com\/preferences\/source?q=https:\/\/www.westernjournal.com\/\">Choose The Western Journal as your preferred source on Google and never miss reporting that defends truth, protects freedom, and advances Western civilization<\/a><\/div>\n<p style=\"border: 1px solid #f5f5f5; padding: 16px;\">Advertise with The Western Journal and reach millions of highly engaged readers, while supporting our work. <a href=\"https:\/\/www.westernjournal.com\/advertise-us\/?wj_source=article\">Advertise Today<\/a>.<\/p>\n<div style=\"position: relative;\">\n<div class=\"ff-fancy-header-container\"> \t\t\t \t<\/div>\n<div class=\"entry-submit-correction inner-content\">\n<div class=\"correction-form\">\n<form style=\"display: none;\">\n<div class=\"sc-name-field\"> \t\t\t\t\t\t<label>* Name<\/label> \t\t\t\t\t\t<br \/> \t\t\t\t\t\t<input type=\"text\" name=\"name\" required> \t\t\t\t\t<\/div>\n<div class=\"sc--field\"> \t\t\t\t\t\t<label>* <\/label> \t\t\t\t\t\t<br \/> \t\t\t\t\t\t<input type=\"text\" name=\"\" required> \t\t\t\t\t<\/div>\n<p> \t\t\t\t\t<label>* Message<\/label> \t\t\t\t\t<br \/> \t\t\t\t\t<textarea name=\"message\" required><\/textarea> \t\t\t\t\t \t\t\t\t\t<\/p>\n<div class=\"required-message\" style=\"display: none; padding-bottom: 15px;\">* All fields are required.<\/div>\n<p> \t\t\t\t\t<input type=\"submit\" value=\"Submit\" onclick=\"event.preventDefault(); firefly_sc();\"> \t\t\t\t\t \t\t\t\t\t<\/p>\n<div class=\"firefly-sc-confirm\" style=\"display: none;\">Success!<\/div>\n<\/p><\/form>\n<\/p><\/div>\n<\/p><\/div>\n<p> \t\t<script> \t\t\tfunction firefly_sc() { \t\t\t\tif( typeof window.captchaPublicKey == typeof undefined ){ \t\t\t\t\tconsole.error('window.captchaPublicKey is not defined'); \t\t\t\t} \t\t\t\tgrecaptcha.execute( window.captchaPublicKey, { action: 'submit_correction' } ).then( function( token ) { \t\t\t\t\tvar opts = { \t\t\t\t\t\taction:    'firefly_sc_submit', \t\t\t\t\t\tname:      document.querySelector( '.entry-submit-correction [name=\"name\"]' ).value, \t\t\t\t\t\t:     document.querySelector( '.entry-submit-correction [name=\"\"]' ).value, \t\t\t\t\t\tmessage:   document.querySelector( '.entry-submit-correction [name=\"message\"]' ).value, \t\t\t\t\t\tpost_id:   firefly_post_id, \t\t\t\t\t\tcap_token: token \t\t\t\t\t}  \t\t\t\t\tvar inputs = [ 'name', '', 'message' ];  \t\t\t\t\tfor( var i = 0; i < inputs.length; i++ ) if( ! ( opts[inputs[i]] = document.querySelector( '.entry-submit-correction [name=\"' + inputs[i] + '\"]' ).value ) ) { \t\t\t\t\t\tdocument.querySelector( '.entry-submit-correction .required-message' ).style.display = 'block'; \t\t\t\t\t\treturn; \t\t\t\t\t}  \t\t\t\t\tdocument.querySelector( '.entry-submit-correction input[type=\"submit\"]' ).style.display = 'none'; \t\t\t\t\tdocument.querySelector( '.entry-submit-correction .firefly-sc-submitting-img' ).src = firefly_loading_gif_url; \t\t\t\t\tdocument.querySelector( '.entry-submit-correction .firefly-sc-submitting-img' ).style.display = 'inline-block';  \t\t\t\t\tconsole.log( 'ma subbing' );  \t\t\t\t\tif( firefly_post_id ) opts['post_id'] = firefly_post_id;  \t\t\t\t\t\/* Send the data using post with element id name and name2*\/ \t\t\t\t\tvar posting = jQuery.post( firefly_ajax_url, opts );  \t\t\t\t\t\/* Alerts the results *\/ \t\t\t\t\tposting.done( function( response ) { \t\t\t\t\t\tif( response.success ) { \t\t\t\t\t\t\tconsole.log( response.data ); \t\t\t\t\t\t\tdocument.querySelector( '.entry-submit-correction .firefly-sc-submitting-img' ).style.display = 'none'; \t\t\t\t\t\t\tdocument.querySelector( '.entry-submit-correction .firefly-sc-confirm' ).style.display = 'block';  \t\t\t\t\t\t\tdataLayer.push( { 'event': 'submit-correction' } ); \t\t\t\t\t\t} \t\t\t\t\t}); \t\t\t\t}); \t\t\t} \t\t<\/script> \t     \t\t\t\t\t\t     \t\t\t\t\t<\/div>\n<p> \t\t\t\t\t \t\t\t<script>console.log(\"ad slot (BA1)\")<\/script> \t\t<\/p>\n<div class=\"ff-fancy-header-container\"> \t\t\t \t<\/div>\n<\/section>\n<p><\/body><\/html><\/p>\n<p><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>OpenAI&#8217;s report details AI model hacking into Hugging Face, exposing credentials<\/p>\n","protected":false},"author":1,"featured_media":2651702,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_mo_disable_npp":"","fifu_image_url":"https:\/\/www.westernjournal.com\/wp-content\/uploads\/2026\/08\/IMG_2091.jpeg","fifu_image_alt":"","footnotes":""},"categories":[33651],"tags":[84852,34803,11208,20164],"class_list":["post-2651701","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-the-western-journal","tag-confidential-info","tag-data-breach","tag-hacking","tag-openai"],"fifu_image_url":"https:\/\/www.westernjournal.com\/wp-content\/uploads\/2026\/08\/IMG_2091.jpeg","_links":{"self":[{"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/posts\/2651701","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/comments?post=2651701"}],"version-history":[{"count":3,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/posts\/2651701\/revisions"}],"predecessor-version":[{"id":2651708,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/posts\/2651701\/revisions\/2651708"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/media\/2651702"}],"wp:attachment":[{"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/media?parent=2651701"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/categories?post=2651701"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.conservativenewsdaily.net\/breaking-news\/wp-json\/wp\/v2\/tags?post=2651701"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}