{"id":214102,"date":"2026-06-24T13:52:11","date_gmt":"2026-06-24T13:52:11","guid":{"rendered":"https:\/\/www.europesays.com\/ro\/214102\/"},"modified":"2026-06-24T13:52:11","modified_gmt":"2026-06-24T13:52:11","slug":"cat-de-repede-a-identificat-mythos-vulnerabilitati-in-sistemele-sua","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ro\/214102\/","title":{"rendered":"C\u00e2t de repede a identificat Mythos vulnerabilit\u0103\u021bi \u00een sistemele SUA"},"content":{"rendered":"

Un model de inteligen\u021b\u0103 artificial\u0103 dezvoltat de compania american\u0103 Anthropic a reu\u0219it s\u0103 identifice vulnerabilit\u0103\u021bi \u00een sisteme informatice clasificate ale guvernului SUA \u00een doar c\u00e2teva ore, potrivit unor oficiali cita\u021bi de Euronews<\/a>. <\/p>\n

Rezultatele au fost ob\u021binute \u00een cadrul unor teste de securitate desf\u0103\u0219urate \u00eempreun\u0103 cu agen\u021bii americane de informa\u021bii, \u00eentr-un exerci\u021biu care ridic\u0103 noi \u00eentreb\u0103ri despre ritmul accelerat al evolu\u021biei AI \u00een zona securit\u0103\u021bii cibernetice.<\/p>\n

– articolul continu\u0103 mai jos –<\/p>\n

Modelul, denumit \u201eMythos\u201d \u0219i dezvoltat de Anthropic, a fost utilizat \u00eentr-o serie de simul\u0103ri menite s\u0103 evalueze capacitatea sistemelor avansate de inteligen\u021b\u0103 artificial\u0103 de a identifica puncte slabe \u00een infrastructuri digitale sensibile.<\/p>\n

Un oficial american a precizat c\u0103 descoperirea vulnerabilit\u0103\u021bilor nu implic\u0103 automat \u0219i exploatarea lor, \u00eens\u0103 viteza cu care AI a reu\u0219it s\u0103 le detecteze a atras aten\u021bia exper\u021bilor \u00een securitate.<\/p>\n

\u201eA intrat \u00een aproape toate sistemele clasificate\u201d<\/p>\n

Subiectul a fost adus \u0219i \u00een aten\u021bia Senatului SUA, unde senatorul democrat Mark Warner a f\u0103cut referire la rezultatele testelor \u00een cadrul unei audieri dedicate securit\u0103\u021bii cibernetice. Potrivit acestuia, informa\u021biile furnizate de Agen\u021bia Na\u021bional\u0103 de Securitate (NSA) \u0219i de Comandamentul Cibernetic american arat\u0103 c\u0103 sistemul testat ar fi reu\u0219it s\u0103 \u201ep\u0103trund\u0103 \u00een aproape toate sistemele clasificate analizate, nu \u00een s\u0103pt\u0103m\u00e2ni, ci \u00een ore\u201d.<\/p>\n

Afirma\u021bia a alimentat \u00eengrijor\u0103rile privind capacitatea noilor modele AI de a automatiza procese care, \u00een mod tradi\u021bional, necesitau echipe \u00eentregi de speciali\u0219ti \u00een securitate informatic\u0103.<\/p>\n

Exerci\u021biile au f\u0103cut parte din ini\u021biativa \u201eProject Glasswing\u201d, lansat\u0103 de Anthropic pentru a evalua impactul modelelor AI de ultim\u0103 genera\u021bie asupra securit\u0103\u021bii na\u021bionale, economiei \u0219i infrastructurii critice. Obiectivul programului este de a \u00een\u021belege at\u00e2t poten\u021bialul defensiv, c\u00e2t \u0219i riscurile asociate utiliz\u0103rii acestor tehnologii \u00een medii sensibile.<\/p>\n

\u00cen paralel, rela\u021bia dintre Anthropic \u0219i administra\u021bia american\u0103 a devenit mai tensionat\u0103 \u00een ultimele luni. De\u0219i compania colaboreaz\u0103 cu autorit\u0103\u021bile \u00een domeniul securit\u0103\u021bii, aceasta \u0219i-a exprimat \u00eengrijorarea cu privire la posibila utilizare militar\u0103 a modelelor sale avansate.<\/p>\n

\u00cen acela\u0219i timp, Washingtonul ar fi impus restric\u021bii privind accesul la anumite sisteme AI dezvoltate de companie, inclusiv limitarea utiliz\u0103rii de c\u0103tre cet\u0103\u021beni str\u0103ini a modelelor \u201eFable 5\u201d \u0219i \u201eMythos 5\u201d.<\/p>\n

Dezbatere \u00een industrie<\/p>\n

M\u0103surile au generat reac\u021bii \u00een comunitatea tehnologic\u0103. Peste 100 de exper\u021bi \u00een securitate cibernetic\u0103 \u0219i reprezentan\u021bi ai industriei au semnat o scrisoare deschis\u0103 \u00een care cer reevaluarea restric\u021biilor. Ace\u0219tia sus\u021bin c\u0103, de\u0219i modelele Anthropic sunt extrem de eficiente \u00een identificarea vulnerabilit\u0103\u021bilor software, astfel de capabilit\u0103\u021bi nu sunt unice \u00een industrie, iar limitarea accesului ar putea afecta eforturile de ap\u0103rare cibernetic\u0103.<\/p>\n

\u00cen opinia semnatarilor, \u00eentr-un context global \u00een care competi\u021bia pentru dezvoltarea AI avansate se intensific\u0103, restric\u021bionarea unor instrumente de securitate ar putea crea dezechilibre strategice.<\/p>\n

Un mesaj public al companiei, distribuit pe platforma X, descria modelul \u201eClaude Fable 5\u201d drept un sistem de clas\u0103 \u201eMythos\u201d, considerat cel mai avansat oferit p\u00e2n\u0103 acum publicului.<\/p>\n

Cazul \u201eMythos\u201d ilustreaz\u0103 astfel schimbarea rapid\u0103 a echilibrului dintre inteligen\u021ba artificial\u0103 \u0219i securitatea cibernetic\u0103: de la un instrument folosit preponderent pentru automatizare \u0219i generare de con\u021binut, AI devine tot mai mult un actor activ \u00een identificarea \u0219i poten\u021bialul management al vulnerabilit\u0103\u021bilor din infrastructurile critice.<\/p>\n","protected":false},"excerpt":{"rendered":"Un model de inteligen\u021b\u0103 artificial\u0103 dezvoltat de compania american\u0103 Anthropic a reu\u0219it s\u0103 identifice vulnerabilit\u0103\u021bi \u00een sisteme informatice…\n","protected":false},"author":2,"featured_media":214103,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[12],"tags":[32,33,31,36,37,27,74,34,35,18479,60043,25,41,40,38,39,60044,26,28,29,30,60045,71,72,73],"class_list":["post-214102","post","type-post","status-publish","format-standard","has-post-thumbnail","category-international","tag-breaking-news","tag-breakingnews","tag-cele-mai-populare-subiecte","tag-featured-news","tag-featurednews","tag-headlines","tag-international","tag-latest-news","tag-latestnews","tag-model-ai","tag-mythos-anthropic","tag-news","tag-ro","tag-romana","tag-romania","tag-romanian","tag-sisteme-clasificate","tag-stiri","tag-titluri","tag-top-stories","tag-topstories","tag-vulnerabilitati-sua","tag-world","tag-world-news","tag-worldnews"],"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@ro\/116805418772506217","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/posts\/214102","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/comments?post=214102"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/posts\/214102\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/media\/214103"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/media?parent=214102"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/categories?post=214102"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ro\/wp-json\/wp\/v2\/tags?post=214102"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}