\n<\/p>\n
Elon Musk<\/a>\u2019\u0131n yapay zeka<\/a> sohbet botu Grok, b\u00fcy\u00fck bir gizlilik tart\u0131\u015fmas\u0131n\u0131n merkezinde.<\/p>\n Google’daki arama sonu\u00e7lar\u0131nda, kullan\u0131c\u0131lar\u0131n bilgisi d\u0131\u015f\u0131nda y\u00fcz binlerce Grok konu\u015fmas\u0131n\u0131n ortaya \u00e7\u0131kt\u0131\u011f\u0131 tespit edildi.<\/p>\n BBC\u2019nin aktard\u0131\u011f\u0131na g\u00f6re, Grok kullan\u0131c\u0131lar\u0131 sohbet d\u00f6k\u00fcmlerini payla\u015fmak i\u00e7in bir butona bast\u0131\u011f\u0131nda, yaln\u0131zca sohbeti g\u00f6nderecekleri hedef ki\u015fiye de\u011fil, ayn\u0131 zamanda internet \u00fczerinde herkese a\u00e7\u0131k \u015fekilde eri\u015filebilen ba\u011flant\u0131lar olu\u015fturuluyor.<\/p>\n Bu ba\u011flant\u0131lar arama motorlar\u0131 taraf\u0131ndan da indekslenmi\u015f durumda.<\/p>\n Sadece Google\u2019da per\u015fembe g\u00fcn\u00fc yap\u0131lan bir aramada neredeyse 300 bin Grok konu\u015fmas\u0131n\u0131n listelendi\u011fi g\u00f6r\u00fcld\u00fc. Forbes ise bu say\u0131n\u0131n 370 bini a\u015ft\u0131\u011f\u0131n\u0131 yazd\u0131.<\/p>\n S\u0131zan konu\u015fmalar aras\u0131nda, kullan\u0131c\u0131n\u0131n Grok\u2019tan g\u00fcvenli \u015fifre olu\u015fturmas\u0131n\u0131 istemesi, kilo vermeye y\u00f6nelik yemek planlar\u0131 talep etmesi, t\u0131bbi durumlarla ilgili ayr\u0131nt\u0131l\u0131 sorular y\u00f6neltmesi gibi g\u00fcndelik \u00f6rnekler yer al\u0131yor.<\/p>\n Ancak BBC\u2019nin haberle\u015ftirdi\u011fi baz\u0131 kay\u0131tlar, chatbot\u2019un yasa d\u0131\u015f\u0131 i\u00e7erikler \u00fcretebildi\u011fini de g\u00f6steriyor. \u00d6rne\u011fin, bir konu\u015fmada Grok\u2019un, laboratuvarda s\u0131n\u0131f A uyu\u015fturucu \u00fcretimine dair talimatlar verdi\u011fi g\u00f6r\u00fcld\u00fc.<\/p>\n Bu olay, yapay zeka sohbet botlar\u0131yla yap\u0131lan g\u00f6r\u00fc\u015fmelerin \u201cpayla\u015f\u201d butonu arac\u0131l\u0131\u011f\u0131yla beklenenden daha geni\u015f kitlelere a\u00e7\u0131ld\u0131\u011f\u0131 ilk \u00f6rnek de\u011fil.<\/p>\n Daha \u00f6nce OpenAI\u2019nin ChatGPT\u2019sinde de payla\u015f\u0131lan konu\u015fmalar arama sonu\u00e7lar\u0131nda g\u00f6r\u00fcnm\u00fc\u015f, \u015firket bunun \u201cdeneysel\u201d bir \u00f6zellik oldu\u011funu a\u00e7\u0131klayarak geri ad\u0131m atm\u0131\u015ft\u0131.<\/p>\n Benzer \u015fekilde, Meta\u2019n\u0131n sohbet botu Meta AI ile yap\u0131lan konu\u015fmalar da uygulamada herkese a\u00e7\u0131k bir \u201cdiscover\u201d ak\u0131\u015f\u0131nda g\u00f6r\u00fcnt\u00fclenmi\u015f ve tepki \u00e7ekmi\u015fti.<\/p>\n Uzmanlar, sohbetlerin anonimle\u015ftirilmi\u015f g\u00f6r\u00fcnse bile kullan\u0131c\u0131lar\u0131n \u00f6zel hayat\u0131na dair hassas bilgiler bar\u0131nd\u0131rabilece\u011fine dikkat \u00e7ekiyor. Oxford \u0130nternet Enstit\u00fcs\u00fc’nden Prof. Luc Rocher, BBC’ye yapt\u0131\u011f\u0131 a\u00e7\u0131klamda, \u201cYapay zeka sohbet robotlar\u0131 gizlilik felaketi olmaya do\u011fru gidiyor\u201d dedi. “Bir kez internete s\u0131zan bu konu\u015fmalar, orada sonsuza kadar kal\u0131r.”300 B\u0130N GROK KONU\u015eMASI<\/strong><\/h3>\n
CHATGPT DE AYNI KR\u0130Z\u0130 YA\u015eAMI\u015eTI<\/strong><\/h3>\n
UZMANLARDAN UYARI<\/strong><\/h3>\n
Rocher\u2019e g\u00f6re bu s\u0131z\u0131nt\u0131larda kullan\u0131c\u0131lar\u0131n isimlerinden konum bilgilerine, ruh sa\u011fl\u0131\u011f\u0131na ili\u015fkin detaylardan i\u015f operasyonlar\u0131na kadar \u00e7ok \u00e7e\u015fitli \u00f6zel veri a\u00e7\u0131\u011fa \u00e7\u0131kabiliyor.<\/p>\n
Grok’u geli\u015ftiren xAI’dan ise hen\u00fcz bir a\u00e7\u0131klama gelmedi.<\/p>\n<\/div>\n