I-OpenAI iphakamisa ukunikeza uhulumeni kaTrump isabelo esingu-5%, kubika i-FT ↗
Kubikwa ukuthi i-OpenAI ixoxe ngokunikeza uhulumeni wase-US isabelo esingu-5%, ngokwengxenye njengoba ingcindezi ikhula mayelana nokuthi ubani ozuzayo ekulinganisweni okukhulu kwe-AI. Kuyifindo elincane elihlelekile lezimali zenqubomgomo.
Lesi siphakamiso ngokusobala sizocindezela nezinye izinkampani ezinkulu ze-AI zase-US ukuba zifake isandla esifanayo kuhulumeni, yize kusekude ukuthi zizokwenza okufanayo. Okwamanje, ukuqaliswa kwakamuva kwemodeli ye-OpenAI sekuvele kubanjwe kunethiwekhi entsha yokuhlolwa yaseWashington.
IMicrosoft yethula inkampani ukusiza izinkampani ukuthi zamukele i-AI ngama-dollar ayizigidi eziyizinkulungwane ezingama-2.5 ↗
I-Microsoft idala i-Microsoft Frontier Company, ibhizinisi elisha lokusabalalisa i-AI elixhaswe ngama-dollar ayizigidi eziyizinkulungwane ezingu-2.5. Umsebenzi wayo: ukusiza izinkampani ezinkulu ukukhetha, ukwenza ngezifiso, nokuthola inani kumathuluzi e-AI - hhayi nje ukuncoma ideshibhodi nokububula.
Ingxenye ethakazelisayo ukuguquguquka kwemodeli. I-Microsoft ithi amakhasimende afuna ukuxuba amathuluzi avela ku-Microsoft kanye nabahlinzeki bangaphandle, bawaxhumanise nedatha yangaphakathi, futhi bagcine imiphumela kunokufaka yonke into ebukhweni bomama.
UZuckerberg weMeta uthi ubuchwepheshe be-AI agent buhamba kancane kunalokho obekulindelwe ↗
UMark Zuckerberg utshele abasebenzi ukuthi ama-ejenti e-AI e-Meta awazange asheshise njengoba bekulindelekile. Lokho kuwukuvuma okucacile, ikakhulukazi ngemuva kokuhlelwa kabusha okukhulu okuhlobene nomsebenzi we-AI kanye nokukhula kokusebenza kahle.
I-Meta isalindele izinzuzo ezinkulu ngokusetshenziswa kwayo kwe-AI, kodwa isimo sengqondo asisincane “sabasebenzi bamarobhothi abasheshayo” kodwa “hmm, umfundi werobhothi usadinga ukugadwa.” Okunye okungajwayelekile: I-Meta ingabuyisa isofthiwe yayo yokulandelela igundane kuphela ngokukhetha ukungena ngemuva kokukhathazeka ngedatha yabasebenzi.
Imodeli entsha, engabizi kakhulu ye-AI yaseShayina ihlangana ne-Anthropic, i-OpenAI endaweni yabo yasekhaya ↗
I-GLM-5.2 ka-Z.ai ithola ukunakwa okukhulu ngoba inikeza ikhodi enamandla nokusebenza kahle kwe-ejensi ngentengo ephansi kakhulu kunamamodeli ahamba phambili asemngceleni wase-US. Ngandlela thile, isikhwama sempahla esingabizi kakhulu sesingene emjahweni wemithwalo kanokusho.
Lo modeli ukhuphuke amazinga onjiniyela kanye namapulatifomu futhi ubhekwa “njengesikhathi esincane se-DeepSeek.” Noma kunjalo, ukwamukelwa kwamabhizinisi aseNtshonalanga kubhekene nokuphazamiseka okukhulu kwesivinini: ukukhathazeka ngokuphepha kwedatha, ikakhulukazi ezimbonini ezilawulwayo.
Imininingwane eyengeziwe mayelana nezivikelo ze-inthanethi ze-Fable 5 kanye nohlaka lwethu lwe-jailbreak ↗
I-Anthropic ishicilele imininingwane eyengeziwe mayelana nezivikelo zokuphepha kwe-inthanethi ze-Fable 5, okuhlanganisa nendlela abahluzi bayo abanquma ngayo ukuthi yini okufanele bayivimbele, bayiqaphe, noma bayivumele. Le nkampani ithi lo modeli uyatholakala emhlabeni wonke futhi, kugxilwe kakhulu ekusetshenzisweni kwe-inthanethi okuyingozi.
Iphinde yaphakamisa uhlaka lobunzima be-cyber jailbreak, ibeka amazinga e-jailbreak ngokukhula kwamandla, ububanzi, ukulula kokusebenzisa izikhali, kanye nokutholakala kalula. Ingabe i-bureaucratic kancane? Yebo. Kuyadingeka? Mhlawumbe, ngoba "isenzo esibi senza into embi" akuyona into ebalulekile ohlelweni lokuphatha.
Idatha ye-Crunchbase: Ukutshalwa Kwezimali Kwezinkampani Eziqalayo Zomhlaba Wonke Kufinyelele Irekhodi Eliyi-$510B Ngo-H1 2026 Njengoba I-AI Boom Isheshisa Ukuxhaswa Kwezimali Nokuphuma ↗
Uxhaso lwezimali lwezinkampani ezisafufusa emhlabeni wonke lufinyelele ku-$510 billion engxenyeni yokuqala, kanti i-AI yenza imali eningi kakhulu. I-OpenAI kanye ne-Anthropic zona ngokwazo zithengise u-$217 billion, okungeyona "indlela yomkhakha" kodwa "izindlovu ezimbili ezihleli kusofa lokuhweba."
Idatha iphinde ibonise imali ye-AI esakazeka ngale kwamalebhu ayisisekelo iye kwingqalasizinda, ezokuvikela, amarobhothi, kanye nokunakekelwa kwempilo. Ngakho-ke yebo, ukuchuma kugxile kakhulu - kodwa kungenela nasemakamelweni aseduze.
Imibuzo Evame Ukubuzwa
Kungani i-OpenAI inganika uhulumeni wase-US isabelo esingu-5%?
Isiphakamiso esibikiwe sibonakala sihlobene nokucindezeleka okukhula ngokuthi obani abazuzayo ekulinganisweni kwezinkampani ze-AI ezinkulu kakhulu. Isabelo sikahulumeni singase sethulwe njengendlela yokwabelana ngobuhle nomphakathi, yize lesi sihloko siphawula ukuthi lo mbono usalokhu ungacacile futhi uyinkimbinkimbi kwezepolitiki. Kungase futhi kwandise ingcindezi kwezinye izinkampani ezinkulu ze-AI zase-US ukuthi zicabangele amalungiselelo afanayo.
Yini okufanele yenziwe yi-Microsoft Frontier Company?
Inkampani iMicrosoft Frontier ichazwa njengebhizinisi elisha lokusabalalisa i-AI elixhaswe ngamaRandi ayizigidi eziyizinkulungwane ezingu-2.5. Indima yalo ukusiza izinkampani ezinkulu ukukhetha, ukwenza ngezifiso, nokuthola inani kumathuluzi e-AI. Ukugcizelelwa kusekusetshenzisweni okusebenzayo kunokuhlola, okuhlanganisa ukuxhuma izinhlelo ze-AI kudatha yangaphakathi namathuluzi asekelayo avela kokubili kubahlinzeki beMicrosoft nabangaphandle.
Kungani ama-ejenti e-AI e-Meta ethuthuka kancane kunalokho obekulindelwe?
Kubikwa ukuthi uMark Zuckerberg utshele abasebenzi ukuthi ubuchwepheshe be-AI agent ye-Meta abuzange busheshise ngokushesha njengoba bekulindelekile. Lesi sihloko sichaza lokhu njengokuvuma okuphawulekayo ngemuva kokuhlelwa kabusha okuhlobene nomsebenzi we-AI kanye nokuzuza kokusebenza kahle. Okubalulekile ukuthi ama-AI agents angase adinge ukugadwa okwengeziwe kanye nokulungiswa ngaphambi kokuba alethe uhlobo lwenzuzo yokukhiqiza izinkampani eziningi ezilindele.
Kungani i-GLM-5.2 ka-Z.ai ibalulekile ezindabeni ze-AI?
I-GLM-5.2 ka-Z.ai idonsela ukunaka ngoba kubikwa ukuthi inikeza ikhodi enamandla kanye nokusebenza kwe-ejenti ngentengo ephansi kakhulu kunezinhlobo ezihamba phambili zase-US. Iphinde yakhuphuka amazinga onjiniyela kanye nepulatifomu futhi iqhathaniswa "nesikhathi esincane se-DeepSeek." Kodwa-ke, ukwamukelwa kwamabhizinisi aseNtshonalanga kungase kukhawulelwe yizinkinga zokuphepha kwedatha, ikakhulukazi ezimbonini ezilawulwayo.
Ziyini izindlela zokuvikela ze-inthanethi ze-Anthropic's Fable 5?
I-Anthropic ichaze ezinye izindlela zokuphepha zokuphepha kwe-inthanethi ze-Fable 5, okuhlanganisa nezindlela zokuhlukanisa ezinquma ukuthi kufanele zivimbele nini, ziqaphe, noma zivumele nini izicelo ezithile ezihlobene ne-inthanethi. Le nkampani iphinde yaphakamisa uhlaka lobunzima be-cyber jailbreak. Lolo hlaka luhlela ama-jailbreak ngezici ezifana nokuzuza kwamandla, ububanzi, ukulula kokusetshenziswa kwezikhali, kanye nokutholakala kalula, okunikeza amaqembu indlela ecacile yokuhlola ukwehluleka okusheshayo okuyingozi.