{"id":120455,"date":"2024-11-02T13:09:07","date_gmt":"2024-11-02T11:09:07","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/google-ispolzoval-bolshuyu-yazykovuyu-model-dlya-vyyavleniya-uyazvimosti-v-sqlite"},"modified":"2024-11-02T13:09:07","modified_gmt":"2024-11-02T11:09:07","slug":"google-ispolzoval-bolshuyu-yazykovuyu-model-dlya-vyyavleniya-uyazvimosti-v-sqlite","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/google-ispolzoval-bolshuyu-yazykovuyu-model-dlya-vyyavleniya-uyazvimosti-v-sqlite","title":{"rendered":"Google p\u00ebrdori nj\u00eb model t\u00eb madh gjuh\u00ebsor p\u00ebr t\u00eb zbuluar nj\u00eb dob\u00ebsi n\u00eb SQLite","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
K\u00ebrkuesit nga nj\u00ebsit\u00eb Google Project Zero dhe Google DeepMind publikuan nj\u00eb raport mbi zhvillimin e sistemit AI Big Sleep, i nd\u00ebrtuar mbi nj\u00eb model t\u00eb madh gjuh\u00ebsor Gemini 1.5 Pro dhe i dedikuar p\u00ebr identifikimin e dob\u00ebsive n\u00eb kodin burimor. Arritja e projektit ishte identifikimi me p\u00ebrdorimin e Big Sleep t\u00eb nj\u00eb dob\u00ebsie t\u00eb njohur m\u00eb par\u00eb si t\u00eb pap\u00ebrdorshme n\u00eb nj\u00eb projekt ekzistues. Dob\u00ebsia u identifikua si rezultat i kontrollit t\u00eb baz\u00ebs s\u00eb kodit t\u00eb DBMS SQLite nga sistemi AI dhe \u00e7on n\u00eb mbushjen e poshtme t\u00eb buffer-it n\u00eb stek (buffer underflow). Problemi u zbulua n\u00eb kodin e pranuar s\u00eb fundmi dhe u eliminua para se t\u00eb shkonte n\u00eb l\u00ebshimin p\u00ebrfundimtar SQLite 3.47.0. <\/p>\n
Modeli mund t\u00eb p\u00ebrdoret si nj\u00eb mjet ndihm\u00ebs n\u00eb fushat ku k\u00ebrkohet nj\u00eb rishikim manual me shum\u00eb pun\u00eb, si dhe p\u00ebr organizimin e verifikimit automatik t\u00eb kodit t\u00eb ri p\u00ebr t\u00eb zbuluar dob\u00ebsi n\u00eb fazat e hershme t\u00eb zhvillimit (para se kodi problematik t\u00eb shkoj\u00eb n\u00eb l\u00ebshimet p\u00ebrfundimtare). Pritet q\u00eb modeli AI i zhvilluar t\u00eb mund\u00ebsoj\u00eb identifikimin e problemeve t\u00eb siguris\u00eb n\u00eb kodin q\u00eb \u00ebsht\u00eb problematik p\u00ebr t'u identifikuar p\u00ebrmes testimit fuzzing. <\/p>\n
Shtes\u00eb mund t\u00eb p\u00ebrmendet deklarata e drejtorit ekzekutiv t\u00eb Google se aktualisht m\u00eb shum\u00eb se nj\u00eb e kat\u00ebrta (25%) e gjith\u00eb kodit t\u00eb krijuar n\u00eb kompani gjenerohet duke p\u00ebrdorur modelet e m\u00ebdha gjuh\u00ebsore Gemini, pas s\u00eb cil\u00ebs ky kod rishikohet dhe pranohet nga inxhinier\u00ebt. V\u00ebrehet se p\u00ebrdorimi i till\u00eb i AI-s\u00eb ka lejuar t\u00eb shpejtohet ndjesh\u00ebm procesi i zhvillimit t\u00eb produkteve. \n Burimi: opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"