{"id":560424,"date":"2026-08-06T04:00:14","date_gmt":"2026-08-06T04:00:14","guid":{"rendered":"https:\/\/www.newsbeep.com\/nz\/560424\/"},"modified":"2026-08-06T04:00:14","modified_gmt":"2026-08-06T04:00:14","slug":"an-ai-model-from-meta-also-hacked-another-company-during-testing","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/nz\/560424\/","title":{"rendered":"An AI model from Meta also hacked another company during testing"},"content":{"rendered":"<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgpz6qy002127p17chsd14y@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            Add Meta to the list of companies with AI agents going rogue. An AI model from the parent company of Facebook and Instagram hacked into another company\u2019s systems during cybersecurity testing, a spokesperson confirmed on Wednesday.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfo000b356r4s6dv0ul@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            Meta says the breach occurred because of an inadvertent error during testing of the model, similar to previously disclosed incidents with OpenAI and Anthropic.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfo000c356r4gip4qv0@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            \u201cA misconfiguration by Irregular, an independent testing company Meta uses, inadvertently allowed one of our models access to the internet during evaluation,\u201d the Meta spokesperson said.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfo000d356ron3cqjf2@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            Meta\u2019s Muse Spark model \u201cexploited a security vulnerability\u201d in another company \u201cin a manner similar to previously-reported instances with other companies.\u201d\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfp000e356r81phwwdm@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            In a statement, Irregular said the incident \u201cis the exact same evaluation-environment issue\u201d that Anthropic <a href=\"https:\/\/www.anthropic.com\/news\/investigating-incidents-cybersecurity-evals\" target=\"_blank\" rel=\"nofollow noopener\">disclosed<\/a> last week that allowed their models access to the open internet before they went on to hack three different organizations\u2019 systems.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfp000f356rtw0v9ek5@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            \u201cThis did not involve a sandbox escape or a sophisticated cyber action. There are no current open issues. Irregular is developing a white paper to share best practices for containment and securely running cyber evals,\u201d the spokesperson added.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfp000g356rzmq3xqk8@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            According to The Information, which first reported on the incident, Meta\u2019s AI model breached an unnamed company\u2019s systems and made changes to its internal system.\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfp000h356rwfp0bf7q@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            Meta says Irregular notified them of the breach, \u201cand we are currently investigating and will issue a full retrospective once we have all the facts.\u201d\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgt3r4600013b6r8nkw10pn@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            A source familiar with the situation told CNN the models have limited internet access in some testing environments to mimic real world threat scenarios, but in this case there was a rare \u201cissue in the setup.\u201d\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgt6yur000d3b6rhmd5ejry@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            \u201cWhat is happening is models are becoming so much more capable, and at the same time evaluations to assess them need to become so much more complex,\u201d the source said. \u201cAnd that just creates room for some mistakes and makes it so that we need to\u2026 up the standards significantly.\u201d\n    <\/p>\n<p class=\"paragraph-elevate inline-placeholder vossi-paragraph_elevate\" data-uri=\"cms.cnn.com\/_components\/paragraph\/instances\/cmsgq2xfp000i356r88cwq6gv@published\" data-editable=\"text\" data-component-name=\"paragraph\" data-article-gutter=\"true\">\n            Meta has now become the third major AI company within a few weeks to disclose an AI model hacking into another company\u2019s systems during testing, highlighting not only the advanced capabilities of AI agents but also some of the potential dangers.\n    <\/p>\n","protected":false},"excerpt":{"rendered":"Add Meta to the list of companies with AI agents going rogue. An AI model from the parent&hellip;\n","protected":false},"author":2,"featured_media":560425,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[365,363,364,111,139,69,145],"class_list":["post-560424","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-new-zealand","tag-newzealand","tag-nz","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/560424","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/comments?post=560424"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/560424\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media\/560425"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media?parent=560424"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/categories?post=560424"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/tags?post=560424"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}