{"id":5261,"date":"2026-07-29T11:00:00","date_gmt":"2026-07-29T11:00:00","guid":{"rendered":"https:\/\/rdgundem.com\/were-running-out-of-reasons-to-ignore-ai-safety\/"},"modified":"2026-07-29T11:00:00","modified_gmt":"2026-07-29T11:00:00","slug":"were-running-out-of-reasons-to-ignore-ai-safety","status":"publish","type":"post","link":"https:\/\/rdgundem.com\/en\/were-running-out-of-reasons-to-ignore-ai-safety\/","title":{"rendered":"We\u2019re running out of reasons to ignore AI safety"},"content":{"rendered":"<figure>\n<p><img decoding=\"async\" alt=\"\" data-caption=\"\" data-portal-copyright=\"Image: The Verge\" data-has-syndication-rights=\"1\" src=\"https:\/\/platform.theverge.com\/wp-content\/uploads\/sites\/2\/2026\/07\/STK485_STK414_AI_SAFETY_B.jpg?quality=90&amp;strip=all&amp;crop=0,0,100,100\"><figcaption>\n\t\t<\/figcaption><\/figure>\n<p class=\"has-drop-cap wp-block-paragraph\">Earlier this month, OpenAI gave several of its AI models a task: complete a <a href=\"https:\/\/arxiv.org\/abs\/2605.11086\">test<\/a> designed to measure their cybersecurity capabilities. It put the systems in a sandboxed environment without an internet connection and set them off to work.<\/p>\n<p class=\"wp-block-paragraph\">What <a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/968988\/openai-hugging-face-hack-ai\">happened next<\/a> is almost laughably silly &#8211; but also, as Adam Gleave, cofounder and CEO of AI safety organization FAR.AI, put it, &#8220;a visceral example of how misaligned AI could cause harm.&#8221; According to OpenAI, the models escaped the sandbox meant to contain them, moved through the company&#8217;s internal systems, found a route to the internet, and then started looking for a way into Hugging Face. And why was \u2026<\/p>\n<p><a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/972380\/open-ai-hugging-face-hack-ai-safety-warning\">Read the full story at The Verge.<\/a><\/p>","protected":false},"excerpt":{"rendered":"<p>Earlier this month, OpenAI gave several of its AI models a task: complete a test designed to measure their cybersecurity capabilities. It put the systems in a sandboxed environment without an internet connection and set them off to work. What happened next is almost laughably silly &#8211; but also, as Adam Gleave, cofounder and CEO &hellip;<\/p>","protected":false},"author":1,"featured_media":5170,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"pagelayer_contact_templates":[],"_pagelayer_content":"","footnotes":""},"categories":[1],"tags":[],"class_list":["post-5261","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-gundem"],"_links":{"self":[{"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/posts\/5261","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/comments?post=5261"}],"version-history":[{"count":0,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/posts\/5261\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/media\/5170"}],"wp:attachment":[{"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/media?parent=5261"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/categories?post=5261"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/rdgundem.com\/en\/wp-json\/wp\/v2\/tags?post=5261"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}