🛡️ I-OpenAI yembula i-GPT-5.4-Cyber ngesonto ngemuva kokumenyezelwa kwemodeli ye-AI yimbangi ↗
I-OpenAI ikhiphe inguqulo elungiselelwe ukuphepha kwe-inthanethi yomodeli wayo oyinhloko, ebizwa ngokuthi i-GPT-5.4-Cyber. Ihloselwe umsebenzi wokuphepha wokuzivikela, ozwakala sengathi uhlosiwe - futhi, kwezinye izindlela, njengokuphendula ngqo emsindweni ozungeze i-Anthropic's Mythos.
Ukuqaliswa kuncane okwamanje. I-OpenAI ithi abathengisi bezokuphepha abahloliwe kuphela, izinhlangano, kanye nabacwaningi abazothola ukufinyelela kuqala, kuyilapho uhlelo lwayo lwe-Trusted Access for Cyber lukhula ngezigaba ezintsha zokuqinisekisa ezivula amakhono aqinile ocwaningo kanye nokuhlaziywa kobungozi. (Reuters)
⚙️ I-Meta yandisa isivumelwano sama-chip enziwe ngokwezifiso ne-Broadcom ukuze iqinise izifiso ze-AI ↗
I-Meta iphinda kabili i-silicon ye-AI eyenziwe ngokwezifiso ne-Broadcom, yandisa ubudlelwano kuze kube ngu-2029. Lesi sivumelwano sihlanganisa ukuzibophezela kokuqala kwamandla angaphezu kwe-gigawatt eyodwa ekhompyutha, okuyinto enkulu kakhulu - enye yalezo zibalo ezizwakala zingaqondakali uze ume kancane ngenxa yobuningi bengqalasizinda eshoyo.
Iphuzu lilula impela: yakha okuningi kwe-stack ngaphakathi endlini futhi uthembele kancane kuma-chip abizayo eNvidia. I-Broadcom izophinde ihlinzeke ngobuchwepheshe bokuxhumana kumaqoqo e-AI e-Meta, kuyilapho i-Meta ithi ama-chip amasha e-MTIA ahlelwe ukuze abhekane nemithwalo yemisebenzi yokucabanga emikhiqizweni yayo. (Reuters)
🧠 Umsunguli we-Anthropic uqinisekisa ukuthi inkampani itshele uhulumeni kaTrump nge-Mythos ↗
I-Anthropic iqinisekisile ukuthi itshele uhulumeni kaTrump nge-Mythos, imodeli yayo enamandla engakakhishwa. UJack Clark uthe uhulumeni udinga ukubonakala ezinhlelweni ezinjengalezi, yize i-Anthropic isalwa noMnyango Wezokuvikela mayelana nokuthola izinkontileka kanye nokufinyelela - inhlanganisela engathandeki, yize lokho kuyisimo senqubomgomo ye-AI njengamanje.
Le nkampani isabambezela ukukhishwa kwe-Mythos emphakathini ngenxa yamandla ayo okuphepha kwe-inthanethi. Ngakho-ke le ndaba ayimayelana kakhulu nokwethulwa komkhiqizo - hhayi ngqo - imayelana nokuthi ubani othola ukufinyelela kusenesikhathi, ubani oxwayiswayo, nokuthi i-AI yasemngceleni isondela kangakanani endaweni yokuphepha kwezwe. (TechCrunch)
🌐 I-Google ingeza amakhono e-AI ku-Chrome ukuze ikusize ulondoloze imisebenzi oyithandayo ↗
I-Google ingeza isici esibizwa ngokuthi Amakhono ku-Chrome, okuvumela abantu ukuthi balondoloze futhi basebenzise kabusha izixwayiso ze-Gemini kuwo wonke amawebhusayithi. Lokho kuzwakala kuncane, mhlawumbe kuyinto ehlekisayo, kodwa kushukumisela isiphequluli ekubeni yindawo yokusebenza ye-AI esikhundleni sokuba nje isitsha sethebhu.
Abasebenzisi bangalondoloza izixwayiso ezivela emlandweni wengxoxo, bazivule ngenkinobho ethi slash noma ethi plus, bese bezisebenzisa ekhasini abalibukayo - kanye namathebhu akhethiwe. I-Google ithi ukusetshenziswa kwasekuqaleni kufaka phakathi ukuqhathanisa ukuthenga, ukulungiswa kwamaresiphi, izifinyezo zedokhumenti, kanye nemisebenzi yezempilo noma yesabelomali, okuzwakala kuwusizo ngomphetho ongaqondakali. (TechCrunch)
🏗️ I-Fluidstack, inkampani entsha ye-AI data center, ixoxisana ngentengo ye-$1B, intengo ye-$18B, ezinyangeni ezimbalwa ngemuva kokufinyelela ku-$7.5B, kusho umbiko ↗
Kubikwa ukuthi i-Fluidstack isezingxoxweni zokuqoqa u-$1 billion ngenani elilinganiselwa ku-$18 billion, okuyizinga elikhulu ukusuka lapho yayilinganiswa khona esikhathini esingeside esidlule. Imakethe isaphatha ingqalasizinda ye-AI njengomoya-mpilo negolide ngesikhathi esifanayo - entulekayo, edingekayo, futhi ebiza kakhulu.
Isizathu esikhulu yisidingo esivela kuma-lab asemngceleni. I-Fluidstack isivele ithole isivumelwano esikhulu ne-Anthropic sokwakha izikhungo zedatha ezigxile ku-AI, futhi lokho kushintshe kusuka ekubeni yinkampani yengqalasizinda ephansi yaba ngelinye lamagama ashisayo ekuxazululeni i-compute. (TechCrunch)
💼 Okukhethekile: Ngaphakathi kwe-Google's AI imisebenzi eqhutshwayo ↗
I-Google izama ukwakha uhlangothi lwemisebenzi lwempikiswano ye-AI ngaphambi kokuba abenzi bomthetho namaqembu ezisebenzi bachaze lona. Isekela izinhlelo ezintsha zocwaningo nokuqeqesha ezigxile endleleni abasebenzi abavumelana ngayo nezimo, okuzwakala kunomusa - futhi kunengqondo.
Lezi zinhlelo zifaka phakathi imizamo yokunakekelwa kwempilo, ukukhiqiza, kanye nokuqeqeshwa, kanye nesinyathelo esisodwa esihlose ukuqeqesha abasebenzi abangu-40,000 ngamakhono e-AI. I-Google iphinde ihlanganise abantu abavela kuhulumeni, embonini, kanye nasemphakathini ukuze baxoxe ngekusasa lomsebenzi, ngakho lokhu kuyingxenye yenqubomgomo yabasebenzi, ingxenye yokuphathwa kwedumela... cishe kokubili. (Axios)
Imibuzo Evame Ukubuzwa
Iyini i-GPT-5.4-Cyber futhi ubani ongayisebenzisa kuqala?
I-GPT-5.4-Cyber iyinguqulo elungiselelwe ukuphepha kwe-inthanethi yemodeli eyinhloko ye-OpenAI, eyakhelwe umsebenzi wokuphepha wokuzivikela. Ukufinyelela kusenesikhathi kukhawulelwe kubathengisi bezokuphepha abahloliwe, izinhlangano, kanye nabacwaningi kunokuba kube ngumphakathi jikelele. Ukuqaliswa kubonakala kuhloselwe ukugcina amakhono anobungozi obukhulu ngaphansi kokulawulwa okuqinile ngenkathi kusasekelwa ucwaningo nokuhlaziywa kobuthakathaka.
Kungani i-OpenAI ikhawulela ukufinyelela kwe-GPT-5.4-Cyber kunokuba iqalise kabanzi?
Lesi sihloko siphakamisa ukuthi i-OpenAI ithatha indlela encane yokukhulula ngoba amamodeli okuphepha kwe-cyber angaba bucayi kakhulu. Ngokwandisa uhlelo lwayo lwe-Trusted Access for Cyber ngezigaba zokuqinisekisa, i-OpenAI inganikeza amathuluzi aqinile kubasebenzisi abagunyaziwe ngenkathi igcina ukuqapha okuseduze. Lokho kwenza ukuqaliswa kuzwakale njengokuthunyelwa kokuphepha okulawulwayo kunokukhishwa komkhiqizo ojwayelekile.
Kungani i-Anthropic's Mythos ibalulekile yize ingatholakali emphakathini?
I-Mythos ibalulekile ngoba ibhekwa njengenamandla ngokwanele ukuqala imihlangano kahulumeni ngaphambi kokukhishwa okubanzi. Lokho kususa ingxoxo ekukhangiseni komkhiqizo iye ekuphepheni kwezwe, ekulawuleni ukufinyelela, nasekubonakaleni kwenqubomgomo. Ngokwezinto ezingokoqobo, kubonisa ukuthi i-AI yasemngceleni iphathwa kanjani njengengqalasizinda yamasu kunokuba nje isofthiwe yezentengiselwano.
Ithini isivumelwano se-Meta se-Broadcom chip mayelana nomjaho wengqalasizinda ye-AI?
Isivumelwano esinwetshiwe se-Meta ne-Broadcom sibonisa ukuthi izinkampani ezinkulu zitshala imali kangakanani ekulawuleni ingxenye enkulu ye-AI stack. Umgomo awukona nje ukwengeza amandla amakhulu okusebenzisa ikhompyutha, kodwa futhi nokunciphisa ukuthembela kuma-chip abizayo e-Nvidia. Kuphinde kubonise ukuthi i-networking kanye ne-inference hardware sekuba yinto ebaluleke kakhulu njengokuthuthukiswa kwemodeli eluhlaza.
Ingakanani i-gigawatt eyodwa yomthamo wokubala we-AI ngokwemibandela engokoqobo?
Lesi sihloko sinikeza i-gigawatt eyodwa njengesibophezelo esikhulu sengqalasizinda, noma ngabe inombolo izwakala ingacacile ekuqaleni. Ngokomongo, ikhomba esikalini sesikhungo sedatha, ukusetshenziswa kwemali yesikhathi eside, kanye nokucindezela okukhulu ukusekela umthwalo wemisebenzi ye-AI ngobuningi obukhulu. Luhlobo lwesibalo olusikisela isifiso sezimboni kunokusetshenziswa kokuhlola.
Isici esisha se-Chrome Skills se-Google sizosisiza kanjani ngempela ekusebenzeni kwansuku zonke?
Kubonakala sengathi i-Chrome Skills yakhelwe ukwenza izikhuthazo ze-Gemini ezigciniwe ziphinde zisetshenziswe kuwo wonke amawebhusayithi ngisho nakuwo wonke amathebhu akhethiwe. Lokho kungenza imisebenzi ejwayelekile isheshe, njengokuqhathanisa imikhiqizo, ukulungisa izindlela zokupheka, ukufingqa amadokhumenti, noma ukuphatha umsebenzi wesabelomali. Ushintsho olukhulu ukuthi i-Chrome iqala ukusebenza njengendawo yokusebenza ye-AI yemisebenzi ephindaphindwayo, hhayi isiphequluli nje kuphela.
Kungani i-Fluidstack ivele yaziswa kakhulu emakethe ye-AI?
Ukwenyuka kwenani elibikiwe le-Fluidstack kubonakala kuhlobene nendlela imakethe evuza ngayo izinkampani ezingahlinzeka ngengqalasizinda ye-AI. Isidingo esivela kuma-lab asemngceleni siyisici esiyinhloko, ikakhulukazi ngemuva kwesivumelwano sayo esikhulu sedatha ne-Anthropic. Ezindaweni eziningi zokucubungula idatha, ukufinyelela ku-computing sekungenye yezithiyo ezibaluleke kakhulu, okusiza ukuchaza iprimiyamu.
Yini i-Google ezama ukuyifeza ngokucindezela kwayo imisebenzi ye-AI?
Kubonakala sengathi i-Google isheshe yaqala ukusebenzisana nohlangothi lwezabasebenzi lwe-AI ukuze ingasiza ekwakheni impikiswano ngaphambi kokuba abenzi bomthetho namaqembu ezisebenzi bayichaze ngaphandle kwabo. Izinhlelo eziphathelene nokunakekelwa kwempilo, ukukhiqiza, ukuqeqeshwa emsebenzini, kanye nokuqeqeshwa kwe-AI ziphakamisa kokubili ukulungiselela abasebenzi kanye nokubeka umphakathi endaweni. Ifundeka njengenhlanganisela yokutshalwa kwezimali kwamakhono, ukuzibandakanya kwenqubomgomo, kanye nesu lodumo.