\n<\/p>\n
OpenAI CEO\u2019su Sam Altman<\/a>, \u015firketin ABD Savunma Bakanl\u0131\u011f\u0131 (Pentagon<\/a>) ile k\u0131sa s\u00fcre \u00f6nce yapt\u0131\u011f\u0131 yapay zeka anla\u015fmas\u0131n\u0131n \u201caceleye getirildi\u011fini\u201d kabul ederek s\u00f6zle\u015fmede de\u011fi\u015fikli\u011fe gidilece\u011fini a\u00e7\u0131klad\u0131.<\/p>\n \u00a0<\/p>\n Altman, sosyal medya platformu X\u2019te payla\u015ft\u0131\u011f\u0131 \u015firket i\u00e7i bir notta, anla\u015fman\u0131n \u00f6zellikle g\u00f6zetleme gibi hassas konularda OpenAI\u2019\u0131n ilkelerini daha a\u00e7\u0131k bi\u00e7imde yans\u0131tacak \u015fekilde g\u00fcncellenece\u011fini belirtti.<\/p>\n \u00a0<\/p>\n Yeni d\u00fczenlemeye g\u00f6re s\u00f6zle\u015fmeye, \u201cyapay zeka sisteminin ABD vatanda\u015flar\u0131na y\u00f6nelik i\u00e7 g\u00f6zetim amac\u0131yla kas\u0131tl\u0131 bi\u00e7imde kullan\u0131lmayaca\u011f\u0131\u201d y\u00f6n\u00fcnde a\u00e7\u0131k bir ifade eklenecek. Notta ayr\u0131ca Savunma Bakanl\u0131\u011f\u0131\u2019n\u0131n, ticari yollarla elde edilen ki\u015fisel veya kimlik belirleyici veriler dahil olmak \u00fczere ABD vatanda\u015flar\u0131n\u0131n kas\u0131tl\u0131 takibi, g\u00f6zetimi veya izlenmesini yasaklayan s\u0131n\u0131rlamay\u0131 kabul etti\u011fi vurguland\u0131.<\/p>\n \u00a0<\/p>\n Notta \u015fu ifadeler de yer al\u0131yor:<\/p>\n \u00a0<\/p>\n \u201cBakanl\u0131k ayr\u0131ca, hizmetlerimizin Sava\u015f Bakanl\u0131\u011f\u0131 istihbarat te\u015fkilatlar\u0131 (\u00f6rne\u011fin NSA) taraf\u0131ndan kullan\u0131lmayaca\u011f\u0131n\u0131 da teyit etti. Bu te\u015fkilatlara y\u00f6nelik herhangi bir hizmet, s\u00f6zle\u015fmemizde ek bir de\u011fi\u015fiklik yap\u0131lmas\u0131n\u0131 gerektirecek.\u201d<\/p>\n \u00a0<\/p>\n Altman a\u00e7\u0131klamas\u0131nda, yapay zeka teknolojisinin hen\u00fcz baz\u0131 alanlar i\u00e7in haz\u0131r olmad\u0131\u011f\u0131n\u0131 belirterek \u015fu ifadeleri kulland\u0131:<\/p>\n \u00a0<\/p>\n \u201cTeknolojinin hen\u00fcz haz\u0131r olmad\u0131\u011f\u0131 pek \u00e7ok konu var ve g\u00fcvenlik a\u00e7\u0131s\u0131ndan hangi \u00f6d\u00fcnlerin verilmesi gerekti\u011fini h\u00e2l\u00e2 tam olarak anlam\u0131\u015f de\u011filiz.\u201d<\/p>\n \u00a0<\/p>\n \u0130\u015f insan\u0131 \u015firketin Pentagon ile teknik g\u00fcvenlik \u00f6nlemleri geli\u015ftirmek i\u00e7in \u00e7al\u0131\u015faca\u011f\u0131n\u0131 da ekledi.<\/p>\n \u00a0<\/p>\n TARTI\u015eMALI ANLA\u015eMA SONRASI GER\u0130 ADIM<\/strong><\/p>\n \u00a0<\/p>\n De\u011fi\u015fiklik karar\u0131, ChatGPT\u2019nin geli\u015ftiricisi OpenAI\u2019\u0131n cuma g\u00fcn\u00fc Pentagon ile yeni bir i\u015f birli\u011fi anla\u015fmas\u0131 yapt\u0131\u011f\u0131n\u0131 duyurmas\u0131n\u0131n ard\u0131ndan geldi.<\/p>\n \u00a0<\/p>\n Anla\u015fma, ABD Ba\u015fkan\u0131 Donald Trump\u2019\u0131n federal kurumlara rakip yapay zeka \u015firketi Anthropic\u2019in ara\u00e7lar\u0131n\u0131 kullanmay\u0131 b\u0131rakma talimat\u0131 vermesinden yaln\u0131zca saatler sonra a\u00e7\u0131klanm\u0131\u015ft\u0131. Ayn\u0131 g\u00fcn Washington y\u00f6netimi \u0130ran\u2019a y\u00f6nelik sald\u0131r\u0131lar ger\u00e7ekle\u015ftirmi\u015fti.<\/p>\n \u00a0<\/p>\n Pentagon ile Anthropic aras\u0131nda \u015firketin yapay zeka modeli Claude\u2019un askeri kullan\u0131m\u0131na ili\u015fkin anla\u015fmazl\u0131k vard\u0131. Anthropic CEO\u2019su Dario Amodei, \u015firketin yapay zeka sistemlerinin \u015fu iki alanda kullan\u0131lmas\u0131na izin vermeyece\u011fini defalarca vurgulam\u0131\u015ft\u0131: ABD i\u00e7inde kitlesel g\u00f6zetim ve insan denetimi olmadan \u00e7al\u0131\u015fan \u00f6l\u00fcmc\u00fcl otonom silah sistemleri.<\/p>\n \u00a0<\/p>\n Pentagon ise yapay zeka sistemlerini \u201ct\u00fcm yasal ama\u00e7lar i\u00e7in\u201d kullanma hakk\u0131na sahip olmas\u0131 gerekti\u011fini savunuyordu.<\/p>\n \u00a0<\/p>\n Claude ara\u00e7lar\u0131n\u0131n Venezuela Devlet Ba\u015fkan\u0131 Nicolas Maduro\u2019nun yakaland\u0131\u011f\u0131 askeri operasyonda da kullan\u0131ld\u0131\u011f\u0131 ortaya \u00e7\u0131km\u0131\u015ft\u0131. Pentagon’un Claude’u veri analiz \u015firketi Palantir ile yap\u0131lan anla\u015fma \u00fczerinden kulland\u0131\u011f\u0131 belirtiliyor.<\/p>\n \u00a0<\/p>\n Wall Street Journal\u2019\u0131n haberine g\u00f6re, ABD Ortado\u011fu Merkez Komutanl\u0131\u011f\u0131 (CENTCOM) da dahil olmak \u00fczere, d\u00fcnyan\u0131n d\u00f6rt bir yan\u0131ndaki komutanl\u0131klar, Anthropic\u2019in yapay zeka ara\u00e7lar\u0131n\u0131 istihbarat analizleri, hedef belirleme ve sava\u015f senaryolar\u0131n\u0131 sim\u00fcle etmek i\u00e7in kulland\u0131.<\/p>\n \u00a0<\/p>\n \u201cACELEYE GELD\u0130\u201d<\/strong><\/p>\n \u00a0<\/p>\n Anthropic ile Pentagon aras\u0131ndaki kriz, yapay zeka sekt\u00f6r\u00fc \u00e7al\u0131\u015fanlar\u0131n\u0131n da tepkisini \u00e7ekmi\u015f, ChatGPT ve Google\u2019dan y\u00fczlerce \u00e7al\u0131\u015fan Anthropic\u2019e destek mektuplar\u0131 imzalam\u0131\u015ft\u0131.<\/p>\n \u00a0<\/p>\n Bu esnada OpenAI\u2019\u0131n Pentagon ile anla\u015fmaya varmas\u0131 da tepkileri alevlendirmi\u015fti.<\/p>\n \u00a0<\/p>\n OpenAI CEO\u2019su, son a\u00e7\u0131klamas\u0131nda, anla\u015fman\u0131n duyurulma bi\u00e7imi konusunda hata yapt\u0131\u011f\u0131n\u0131 kabul etti. Altman payla\u015f\u0131m\u0131nda, \u201cAnla\u015fmay\u0131 cuma g\u00fcn\u00fc duyurmak i\u00e7in acele etmemeliydim\u201d diyerek \u015fu de\u011ferlendirmeyi yapt\u0131:<\/p>\n \u00a0<\/p>\n \u201cGer\u00e7ekten gerilimi d\u00fc\u015f\u00fcrmeye ve daha k\u00f6t\u00fc bir sonucu \u00f6nlemeye \u00e7al\u0131\u015f\u0131yorduk, ancak d\u0131\u015far\u0131dan f\u0131rsat\u00e7\u0131 ve da\u011f\u0131n\u0131k g\u00f6r\u00fcnd\u00fc\u011f\u00fcn\u00fc d\u00fc\u015f\u00fcn\u00fcyorum.\u201d<\/p>\n \u00a0<\/p>\n Altman daha \u00f6nce \u00e7al\u0131\u015fanlara g\u00f6nderdi\u011fi bir notta OpenAI\u2019\u0131n da Anthropic ile ayn\u0131 \u201ck\u0131rm\u0131z\u0131 \u00e7izgilere\u201d sahip oldu\u011funu s\u00f6ylemi\u015fti.<\/p>\n \u00a0<\/p>\n KULLANICILAR BOYKOT ETT\u0130<\/strong><\/p>\n \u00a0<\/p>\n Anla\u015fman\u0131n zamanlamas\u0131 sosyal medyada yo\u011fun ele\u015ftirilere yol a\u00e7m\u0131\u015ft\u0131. \u00c7ok say\u0131da kullan\u0131c\u0131 uygulama ma\u011fazalar\u0131nda ChatGPT\u2019yi b\u0131rak\u0131p Anthropic\u2019in Claude sistemine y\u00f6neldi\u011fini a\u00e7\u0131kl\u0131yor.<\/p>\n \u00a0<\/p>\n Pentagon son gerilimin ard\u0131ndan Anthropic\u2019i \u201ctedarik zinciri riski\u201d olarak s\u0131n\u0131fland\u0131r\u0131lacak. Bu s\u0131n\u0131fland\u0131rma Anthropic’in, ABD’de adeta yabanc\u0131 \u015firket muamelesi g\u00f6rmesine neden olabilir. Bu etiket nedeniyle, ABD h\u00fck\u00fcmetiyle i\u015f yapan t\u00fcm \u015firketlerin Anthropic ile ba\u011flar\u0131n\u0131 kesmesini gerekecek.<\/p>\n \u00a0<\/p>\n Altman ise tart\u0131\u015fmalara ili\u015fkin hafta sonu yapt\u0131\u011f\u0131 g\u00f6r\u00fc\u015fmelerde Anthropic\u2019in \u201ctedarik zinciri riski\u201d ilan edilmemesi gerekti\u011fini savundu\u011funu belirterek, Savunma Bakanl\u0131\u011f\u0131\u2019n\u0131n rakip \u015firkete de OpenAI ile ayn\u0131 \u015fartlar\u0131 sunmas\u0131n\u0131 umdu\u011funu s\u00f6yledi.<\/p>\n \u00a0<\/p>\n Anthropic, 2021 y\u0131l\u0131nda OpenAI\u2019dan ayr\u0131lan eski \u00e7al\u0131\u015fanlar ve ara\u015ft\u0131rmac\u0131lar taraf\u0131ndan kurulmu\u015f ve kendisini \u201c\u00f6nce g\u00fcvenlik\u201d yakla\u015f\u0131m\u0131n\u0131 benimseyen bir alternatif olarak konumland\u0131rm\u0131\u015ft\u0131.<\/p>\n \u00a0<\/p>\n<\/div>\n