{"id":167916,"date":"2026-04-08T11:12:35","date_gmt":"2026-04-08T09:12:39","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/anthropic-otkryla-ogranichennyj-dostup-k-modeli-claude-mythos-preview-ona-avtonomno-ishhet-dyry-v-po-i-uzhe-nashla-tysyachi-uyazvimostej"},"modified":"2026-04-10T10:23:35","modified_gmt":"2026-04-10T08:23:35","slug":"anthropic-otkryla-ogranichennyj-dostup-k-modeli-claude-mythos-preview-ona-avtonomno-ishhet-dyry-v-po-i-uzhe-nashla-tysyachi-uyazvimostej","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/anthropic-otkryla-ogranichennyj-dostup-k-modeli-claude-mythos-preview-ona-avtonomno-ishhet-dyry-v-po-i-uzhe-nashla-tysyachi-uyazvimostej","title":{"rendered":"Anthropic hapi qasjen e kufizuar n\u00eb modelin e saj Claude Mythos Preview \u2014 ai autonomisht k\u00ebrkon vrima n\u00eb software dhe ka zbuluar tashm\u00eb mij\u00ebra dob\u00ebsi.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Kompania Anthropic hapi akses t\u00eb kufizuar n\u00eb modelin e saj t\u00eb ri t\u00eb AI, Mythos, i cili \u00ebsht\u00eb nd\u00ebrtuar p\u00ebr t\u00eb k\u00ebrkuar dob\u00ebsi kritike dhe p\u00ebr t\u00eb reflektuar k\u00ebrc\u00ebnimet kibernetike. Ky mjet do t\u00eb jet\u00eb i disponuesh\u00ebm vet\u00ebm p\u00ebr disa korporata t\u00eb p\u00ebrzgjedhura, sepse aft\u00ebsit\u00eb e tij p\u00ebr t\u00eb hakuar sisteme jan\u00eb konsideruar si shum\u00eb t\u00eb rrezikshme p\u00ebr publikimin e p\u00ebrgjithsh\u00ebm.<\/p>\n

<\/center>\"Anthropic<\/p>\n

Sipas raportin<\/a> The Wall Street Journal, nj\u00eb version paraprak i rrjetit neural do t\u00eb marrin rreth 50 kompani dhe organizata q\u00eb mb\u00ebshtesin infrastruktur\u00ebn kritike t\u00eb IT-s\u00eb, p\u00ebrfshir\u00eb Amazon, Apple, Google dhe Linux Foundation. Sipas drejtorit t\u00eb divizionit Frontier Red Team n\u00eb Anthropic, Logan Graham, modeli i AI Mythos ka treguar aft\u00ebsi kaq t\u00eb larta n\u00eb gjetjen dhe shfryt\u00ebzimin e gabimeve n\u00eb softuer, sa q\u00eb zhvilluesit aktualisht nuk kan\u00eb n\u00eb plan ta hapin at\u00eb p\u00ebr publikun e gjer\u00eb.<\/p>\n

Ekspert\u00ebt e siguris\u00eb kibernetike jan\u00eb gjithnj\u00eb e m\u00eb t\u00eb shqet\u00ebsuar se inteligjenca artificiale po e shkurton fuqish\u00ebm intervalin e koh\u00ebs midis zbulimit t\u00eb dob\u00ebsive dhe krijimit t\u00eb nj\u00eb softueri funksional p\u00ebr shfryt\u00ebzimin e tyre. N\u00eb fund t\u00eb vitit t\u00eb kaluar, studiuesit nga Universiteti i Stanfordit zbuluan se rrjetet neurale k\u00ebrkojn\u00eb dhe p\u00ebrdorin defektet n\u00eb rrjetet reale pothuajse po aq efektivisht sa specialist\u00ebt njer\u00ebzor\u00eb. Nj\u00eb shembull i mir\u00eb \u00ebsht\u00eb modeli i m\u00ebparsh\u00ebm i Anthropic, Claude Opus 4.6, i cili n\u00eb fillim t\u00eb k\u00ebtij viti, vet\u00ebm brenda dy jav\u00ebve, gjeti m\u00eb shum\u00eb dob\u00ebsi kritike n\u00eb shfletuesin Firefox sesa zakonisht zbulohen n\u00eb t\u00eb gjith\u00eb bot\u00ebn p\u00ebr dy muaj.<\/p>\n

Modeli i ri Mythos u tregua rreth 10 her\u00eb m\u00eb efektiv se sistemet e m\u00ebparshme t\u00eb AI n\u00eb llogaritjen e kostos p\u00ebr zbulimin e nj\u00eb gabimi. Megjith\u00ebse Anthropic aktualisht nuk ka n\u00eb plan ta l\u00ebshoj\u00eb produktin e saj n\u00eb akses t\u00eb hapur, Logan Graham \u00ebsht\u00eb i sigurt se n\u00eb vitet e ardhshme do t\u00eb ket\u00eb mund\u00ebsi t\u00eb ngjashme haker\u00ebsh edhe n\u00eb rrjete t\u00eb tjera neurale. N\u00eb k\u00ebt\u00eb lidhje, eksperti i b\u00ebn thirrje komunitetit IT q\u00eb tani t\u00eb p\u00ebrgatitet p\u00ebr realitetin, ku vonesa midis zbulimit t\u00eb nj\u00eb dob\u00ebsie t\u00eb softuerit dhe aplikimit t\u00eb saj n\u00eb nj\u00eb sulm do t\u00eb reduktohet n\u00eb zero.<\/p>\n

Burimi:<\/strong><\/p>\n