Amfani da AI a tsakanin kamfanonin kuɗi na UK da aka yi wa binciken ya kai 86%, amma alƙaluman da suka fi muhimmanci suna ƙarƙashin kanun labarin: ɓangare na uku (third parties) yanzu suna samar da fiye da rabin tsarin high-materiality, kuma sa ido a kansu shi ne ƙalubalen gudanarwa (governance) da kamfanoni suka fi ambata.
Taƙaitawa ga Shugabanni
- 86% na waɗanda suka amsa sun ce suna amfani da AI, sama da 75% a 2024. Generative AI ya kai 56% na tsarin da aka bayar da rahotonsu.
- Kusan 7% na tsarin da aka bayar da rahotonsu aka tantance a matsayin high materiality, ƙasa daga 16% a 2024, duk da haka jimillar adadin tsarin high-materiality ya fi yawa a 2026.
- Third-party implementations sun kai 53% na tsarin high-materiality, sama da 35% a 2024. Manyan masu samar da cloud/compute uku sun kai 79% na dangantakar cloud/compute da aka bayar da rahotonsu.
- 80% na waɗanda suka amsa sun tura AI agents, amma 2% kaɗai sun ce suna da agents masu broad autonomy.
- Waɗanda suka amsa suna sa ran fa'idoji da haɗari duka su tashi a dukkan rukunoni, inda ƙaruwar haɗarin cyber (cyber risk amplification) da dogaro mai muhimmanci ga ɓangare na uku (critical third-party dependencies) suka zama manyan haɗarin tsarin baki ɗaya (systemic risks).
- Sa ido kan tsarin AI na ɓangare na uku shi ne ƙalubalen gudanarwa da aka fi ambata.
Menene binciken, kuma me ba shi ba #
Wallafar, wato bincike na huɗu na Bank of England da FCA kan AI da machine learning a fannin hada-hadar kuɗi na UK, taƙaitaccen bayani ne na jimillar sakamakon binciken, kuma ya dace a riƙe wannan tsari a zuciya kafin a ciro kowane ƙarshe daga gare shi. Sakamakon shaida ne daga kamfanonin da suka shiga, ba cikakken kimantawar karɓuwar AI a dukkan fannin kuɗi na UK ba. Kada a karanta komai a ƙasa a matsayin ƙidaya ta kowa (census).
Binciken ya samu amsoshi 196: manyan kamfanoni 101 da ƙananan kamfanoni 95. A karon farko, kamfanonin da ke da ma'aikata ƙasa da 100 za su iya zaɓar gajeriyar takardar tambayoyi, kuma waɗanda suka amsa 105 sun cika cikakken binciken. Sai dai idan an faɗi akasin haka, alƙaluman rahoton sun dogara ne kan amsoshin cikakken binciken, wanda ke sauƙaƙa kwatance da binciken baya. Ana lissafa kaso ne da ingantattun amsoshi na kowace tambaya kaɗai, don haka kamfanonin da ba su amsa wata tambaya ba ba sa cikin denominator na wannan tambayar. Amsoshin bankunan ƙasa da ƙasa da broker/dealers an bayar da su ne dangane da ayyukansu na UK.
Binciken 2026 ya riƙe tambayoyin ginshiƙi daga binciken baya don ci gaba, yayin da ya ƙara ko ya gyara tambayoyi don nuna ci gaban karɓuwa da ƙarfin AI tun 2024. Shi ya sa rahoton zai iya ajiye alƙaluman 2026 kusa da na 2024 kan karɓuwa, materiality da kason ɓangare na uku, kuma shi ya sa rahoton ke roƙon masu karatu su fassara kwatance da 2024 da taka-tsantsan inda kalmomin tambayoyi, ma'anoni, iyaka ko rukunin waɗanda suka amsa suka bambanta. Wasu batutuwa, kamar gajeren annex kan frontier AI da haɗarin cyber, ba su da kwatance da 2024 kwata-kwata.
Waɗannan gargaɗi suna da muhimmanci saboda da yawa daga cikin alƙaluma mafiya ban mamaki a ƙasa kaso ne na kaso. Mai karatu da ya yi watsi da denominators zai ɗauki sauyin haɗi a matsayin sauyin yawa. Binciken da kansa yana ba da misali bayyananne na wannan tarko, wanda za mu iso gare shi nan ba da jimawa ba.
Karɓuwa ta kusa zama ta kowa, kuma generative AI ne ke kan gaba #
Fara da kanun labarin. 86% na waɗanda suka amsa sun ce suna amfani da AI, sama da 75% a 2024. Wannan alƙalami ya ƙunshi cikakken binciken da gajeren binciken (short-form) duka, kuma rahoton da kansa ya yi gargaɗin cewa kason 2026 da na 2024 ba sa nufin rukunin kamfanoni ɗaya, don haka a kwatanta su da taka-tsantsan. Kamfanoni na ƙara ba da rahoton amfani da generative AI, agents masu amfani da AI da third-party implementations na tsarin AI, kuma waɗannan zaruruwa uku suna ratsa dukkan sauran binciken.
Generative AI ya kai 56% na dukkan tsarin AI da aka bayar da rahotonsu. Kwatancen da rahoton ya yi shi ne da foundation models, waɗanda suka kai 17% na use cases na AI da aka bayar da rahotonsu a 2024. Rahoton ya ce waɗannan ma'aunan ba za a iya kwatanta su kai tsaye ba, don haka kuskure ne a kira shi tsallen maki 39 a fasaha guda; idan aka haɗa su, a kalmomin rahoton, suna nuna ƙaruwa mai yawa a amfanin kamfanoni na ƙarfin general-purpose tun 2024.
Ina AI yake a cikin kamfanoni? Operations da IT sun kai kusan 31% na dukkan tsarin AI da aka bayar da rahotonsu, sama da kusan 22% a 2024, kuma babu wani fanni guda da ya wuce 11%. Wannan taruwa ce mai ban mamaki. Amfanin da ke fuskantar abokan ciniki da na front-office su ne ke samun mafi yawan hankalin jama'a, amma babban rukunin tsarin da aka bayar da rahotonsu yana cikin bututun cikin gida na kasuwancin. Wani alƙalami mai alaƙa daga sashen ƙananan kamfanoni yana nuna alƙibla ɗaya: a cikin waɗanda suka amsa gajeren binciken, operations shi ne fannin da aka fi ambata don amfanin AI na yanzu ko wanda aka tsara, wanda 64% suka ambata.
Fa'idojin da ake gani sun yi daidai da wannan hoto. Ingancin aiki, automation da productivity su ne fannonin fa'ida mafiya girma da ake gani daga AI a yau. Bisa shaidar abin da suke ba da rahotonsa, kamfanoni suna amfani da AI ne galibi don gudanar da kansu cikin inganci.
Karɓuwa ma ba ta bazu daidai ba. Kusan rabin kamfanoni sun ce suna da tsarin AI goma ko ƙasa da haka, yayin da kamfanoni goma da suka fi yawan tsarin suka kai 43% na dukkan tsarin AI da aka bayar da rahotonsu. Don haka alƙalamin 86% yana bayyana faɗin amfani ne, ba zurfinsa ba. Ƙaramin adadin kamfanoni ne ke ɗaukar babban kaso na jimillar, kuma yawancin kamfanoni suna gudanar da tsarin kaɗan kawai. Ƙananan kamfanoni ma suna ciki: 73% na waɗanda suka amsa gajeren binciken sun ce suna amfani da AI, kuma daga cikinsu, 58% sun ce galibi suna amfani da general-purpose large language models.
Ƙalubalen materiality #
Ga tarkon da aka ambata a baya. Kusan 7% na tsarin AI da aka bayar da rahotonsu waɗanda suka amsa sun tantance su a matsayin high materiality, idan aka kwatanta da 16% a 2024. Idan aka karanta da sauri, yana kama da AI yana rage muhimmanci ga kamfanoni, ko yana rage haɗari.
Ba haka yake nufi ba. Jimillar adadin tsarin AI na high-materiality da aka bayar da rahotonsu a 2026 ya fi na 2024. Kason ya faɗi ne saboda jimillar ta girma da sauri fiye da rukunin high-materiality. Yayin da kamfanoni ke tura ƙarin tsarin, yawancin ƙarin ana tantance su a matsayin ƙananan materiality, wanda ke rage kason yayin da adadin tsarin high-materiality ke ci gaba da tashi. Waɗannan tsarin ma sun taru: kamfanoni goma sun kai 62% na dukkan tsarin AI na high-materiality da aka bayar da rahotonsu.
Wannan gyara ne mai amfani ga duk mai gina hujjar gudanarwa bisa kaso kawai. Faɗuwar kason tsarin high-materiality ba ta faɗi komai mai kwantar da hankali game da nauyin aikin mutanen da ke da alhakinsu ba. Akwai ƙarin su da za a sa ido a kai fiye da yadda suke a 2024.
Daga ina tsarin masu muhimmanci ke zuwa #
Sakamako mafi tasiri a binciken shi ne wanda ke ginawa da gudanar da tsarin da suka fi muhimmanci. Third-party implementations sun kai 39% na dukkan tsarin AI da aka bayar da rahotonsu, sama da 33% a 2024 da 17% a 2022. Wannan motsi ne a hankali. Sauyin a cikin tsarin high-materiality ya fi kaifi sosai: third-party implementations sun kai 53% daga cikinsu, sama da 35% a 2024.
Don haka, a cikin dukkan tsarin da aka bayar da rahotonsu, waɗanda kamfanoni ke ɗauka mafiya muhimmanci yanzu sun fi yiwuwa su fito daga waje. Tashin maki 18 cikin shekaru biyu, a rukunin da gazawa za ta fi cutarwa, shi ne alama mafi bayyana a bayanan. Idan aka haɗa da sakamakon materiality na sama, hoton shi ne na ƙaruwar adadin tsarin masu muhimmanci, wanda yawancinsu kamfanoni ba sa gina su da kansu.
Taruwar masu samarwa tana ƙara kaifafa shi. Manyan masu samar da cloud da compute uku sun kai 79% na dukkan dangantakar cloud/compute da aka bayar da rahotonsu, kuma 98% na waɗanda suka amsa sun ce suna amfani da ɗaya daga cikin waɗannan ukun. Samar da model API da hosting ma ya taru, ko da yake ƙasa da haka: manyan masu samarwa uku sun kai 54% na dangantakar da aka bayar da rahotonsu, kuma manyan masu samar da cloud/compute ukun duk suna cikin manyan masu samar da model API/hosting biyar. Kamfanoni na iya samun tsarin AI da yawa, amma suna ɗiba daga ƙaramin rukunin masu samarwa don ababen more rayuwa da models da ke ƙarƙashinsu.
Maye gurbi (substitution) ita ce tambaya mafi wuya. Daga cikin dangantakar cloud da compute daban-daban 169 da aka bayar da rahotonsu, 37% an ɗauke su masu muhimmanci sosai (high criticality) kuma masu wuyar maye gurbi, kuma 35% kaɗai na wannan rukunin ke da shirin fita (exit plans) da aka gwada. Ga dangantakar model API/hosting 226, kason high-criticality da wuyar maye gurbi ya kai 14%. Ƙarin tsammanin operational resilience ga AI na ɓangare na uku, kamar gwajin abubuwan da suka faru tare (joint incident testing) ko tsara fita, 27% kaɗai na kamfanoni suka ambata.
Waɗanda suka amsa da kansu sun taso da sigar wannan damuwa ta tsarin baki ɗaya: cyber risk amplification da critical third-party dependencies su ne systemic risks biyu mafiya girma, a yau da kuma cikin shekaru uku masu zuwa, kuma waɗanda suka amsa suna sa ran su tsananta yayin da amfani da AI ke ƙara yaɗuwa. Kason masu ɗaukar critical third-party dependencies a matsayin damuwa babba ko mai tsanani ya tashi daga 40% a yau zuwa 63% cikin shekara ɗaya zuwa uku masu zuwa. Wannan tsammani hasashe ne na waɗanda suka amsa, ba sakamakon da aka gani ba, amma alƙaluman taruwar masu samarwa suna ba shi tushe bayyananne.
Kamfanoni ba su jahilci takurawar ba. Damuwa kan tsaron bayanai (information security) da operational resilience 40% na waɗanda suka amsa sun gano su a matsayin babbar takura ga karɓar AI na ɓangare na uku, sai ƙarancin bayyana gaskiya (transparency) cikin tsarin AI na ɓangare na uku (bayanan horo, ƙirar model, iyakokin aiki, hanyoyin gazawa) a 34%. Idan aka ƙirga manyan takurawa da matsakaita tare, wahalar tattaunawar kariyar kwangila ita ce ƙalubale na biyu da aka fi ambata. Idan aka haɗa su, suna bayyana kamfanonin da ke dogara sosai ga masu samarwa na waje yayin da suke cewa ba koyaushe ne suke iya ganin cikinsu ba.
Agents: an tura su ko'ina, galibi a matsayin masu taimako #
Alƙaluman agents suna buƙatar karatu da hankali, saboda kanun labarin da ainihin abin suna nuna alƙibla daban-daban. 80% na waɗanda suka amsa sun ce sun tura AI agents, yayin da ƙarin 14% suka ce suna shirin tura su cikin watanni 12 masu zuwa. Bisa waɗannan alƙaluma, ƙaramin ragowa ne kaɗai bai tura ba kuma bai tsara ba.
Amma duba abin da agents ke yi. 80% na waɗanda suka amsa sun ce suna da agents a production da ke ba da decision support, wanda aka ayyana a matsayin AI agents da ke ba da shawarar matakai yayin da mutane ke yanke shawara da aiwatarwa. 46% sun ce suna da agents masu constrained execution, masu aiki cikin iyakoki da ƙa'idoji da aka riga aka ƙayyade, kuma 22% sun ce suna da agents masu conditional autonomy. 2% kaɗai sun ce sun tura agents masu broad autonomy. Babban tsarin shi ne agents da ke ba da shawara ko suke aiki cikin ƙuntatattun iyakoki.
Sadarwa da abokan ciniki ita ce inda layin ya fara motsawa. 16% sun ce suna amfani da agents a yanzu don sadarwa da abokan ciniki ta atomatik, ya kai 34% idan aka haɗa da turawar da aka tsara. Wannan ya ninka kason fiye da sau biyu idan kamfanoni suka cika tsare-tsarensu, ko da yake tsare-tsare tsare-tsare ne, kuma binciken ba zai iya gaya mana nawa za a aiwatar ba. Akasin haka, ko da an haɗa da turawar da aka tsara, bai fi 9% na waɗanda suka amsa ba ne ke sa ran agents su amince ko su ƙi sakamakon abokan ciniki, su aiwatar da ciniki ko su sanya oda, su fara, su aiwatar ko su toshe biyan kuɗi, ko su sauya iyakoki, ma'aunai da matakan kariya.
Fassarata ita ce alƙalamin 2% ya fi bayyana abubuwa, kuma ya cancanci hankali fiye da 80%. Tura agent da ke ba da shawara sannan ya jira mutum mataki ne matsakaici. Tura wanda ke da broad autonomy yana canza wanda ke da alhakin sakamako. Kamfanoni ya zuwa yanzu sun yarda su ɗauki mataki na farko ko'ina, na biyun kuma kusan babu. Ko tsare-tsaren sadarwa da abokan ciniki alama ce ta farkon sauyi a wannan matsayi na ɗaya daga cikin tambayoyi masu ban sha'awa da ke buɗe a bayanan.
Ana sa ran fa'idoji da haɗari duka su tashi #
Waɗanda suka amsa ba sa ganin AI a matsayin caca mai gefe ɗaya. Suna sa ran fa'idoji su ƙaru a dukkan fannoni cikin shekaru uku masu zuwa, inda ƙaruwa mafiya girma da ake tsammani ke cikin cybersecurity da operational resilience, da samar da kuɗaɗen shiga da matsayin gasa. Waɗannan tsammani ne, kuma binciken ya gabatar da su haka. Babu ɗaya daga cikin fa'idojin da aka lissafa da aka ɗauka ko'ina a matsayin wanda ya wuce matsakaici a girma a yau.
Gefen haɗari ma yana da daidaito, ta fuskar alƙibla. Fannonin haɗari mafiya girma da ake gani ga kamfanoni daga AI a yau su ne kariyar bayanai da tsaro, haɗarin ɓangare na uku da supply chain, da haɗarin musamman na GenAI. Cikin shekara ɗaya zuwa uku masu zuwa, waɗanda suka amsa suna sa ran matakan haɗari su ƙaru a dukkan rukunoni, ko da yake a matsakaici sun ɗauki haɗari a matsayin marasa tsanani a 2026 fiye da 2024.
Tushen ya tsara hoton gaba ɗaya cikin daidaito. Waɗannan ci gaba za su iya kawo fa'idoji kamar ƙarin inganci, ingantaccen tallafin yanke shawara da ƙarfafa sarrafa haɗari. Za su iya kuma haifar ko ƙara haɗari da suka shafi gudanarwa, operational resilience, kariyar bayanai, sakamakon masu amfani, dogaro ga ɓangare na uku da daidaiton harkokin kuɗi. Sassaucin na tushen ne, kuma ya dace; binciken yana auna fahimta da al'adar da aka bayar da rahotonta, ba riba ko asarar da ta tabbata ba.
Wani rukunin sakamako ya shafi frontier AI da ƙarfin cyber. 88% na waɗanda suka amsa sun tantance ƙarfin cyber na frontier AI da ke canzawa cikin shekaru biyu masu zuwa a matsayin mai tasiri matsakaici ko babba kan risk profile na kamfaninsu. Wannan kimantawa ce ta gaba maimakon tasirin da aka auna, ko da yake 19% tuni sun ambaci ƙaruwar tasirin hare-haren supply chain ko na ɓangare na uku a matsayin hanyar da frontier AI ke haifar da haɗari mafi girma a yanzu. Abin da kamfanoni suka riga suka yi a martani ya fi sauƙin tabbatarwa: 64% sun zaɓi vulnerability management da saurin patching a matsayin fannin cyber resilience da ya samu ƙaruwar hankalin shugabanni mafi girma sakamakon sanin frontier AI.
Takurawa: ta aiki fiye da ta ƙa'idoji #
Idan ka yi tsammanin ƙa'idoji (regulation) za su kasance a saman jerin cikas, binciken ya fi zurfi. Takurawa mafiya ambato da ba na ƙa'idoji ba su ne wahalar haɗa AI cikin hanyoyin kasuwanci da yanke shawara da ake da su, da tsofaffin tsarin (legacy systems) ko ababen more rayuwa, waɗanda 23% da 22% na waɗanda suka amsa suka gano a matsayin manyan takurawa bi da bi. Waɗannan kaso ne matsakaita, wanda ke nuna babu wani cikas guda na waje da ƙa'idoji da ya mamaye, amma matsalolin injiniya da na tsarin ƙungiya ne na yau da kullum, ba na ban mamaki ba.
A gefen ƙa'idoji, takurawa mafiya ambato da ake gani su ne kariyar bayanai da sirri na UK, da ƙa'idojin ƙasa da ƙasa, waɗanda 42% da 29% na waɗanda suka amsa bi da bi suka gani a matsayin nauyin bin ƙa'ida mai girma. 28% sun nuna babu takurawar ƙa'idoji.
Ƙananan kamfanoni suna nuna tsari makamancin haka: 30% na waɗanda suka amsa gajeren binciken sun gano kariyar bayanai, sirri da batutuwan IT a matsayin takurawa. Kariyar bayanai tana bayyana a ƙimar haɗari, alƙaluman nauyin ƙa'idoji da takurawar ƙananan kamfanoni. Idan akwai jigo guda da ke haɗa sassan binciken daban-daban baya ga ɓangare na uku, to wannan ne.
Gudanarwa: faɗi a takarda, sirara bayan turawa #
Tsare-tsaren hukuma sun kusa zama na kowa. 94% na waɗanda suka amsa sun ce suna da manufofi, ƙa'idodi ko matsayin AI na dukkan kamfani, kuma 84% sun ce suna da dandalin gudanarwar AI/model-risk na tsakiya (centralised governance fora). Daga cikin matakan kariya na haɗarin bayanai, amincewar ɗan adam, bita kafin turawa (pre-deployment review) da vendor due diligence kowanne fiye da 90% na waɗanda suka amsa sun ambata.
Yanzu kwatanta matakan kariya da ke aiki bayan ƙaddamarwa. Ongoing monitoring, gwajin zubewar bayanai (leakage testing) da bita na lokaci-lokaci ko sabunta takardar shaida (recertification) kowanne kusan rabin waɗanda suka amsa ne suka ambata. Bambancin a sarari yake. Matakan kariya na haɗarin bayanai da ke faruwa kafin ko a lokacin turawa, inda mutum ke sa hannun amincewa, sun kusa zama na kowa. Waɗanda ke ci gaba da aiki bayan turawa kusan rabi ne ke ambatonsu.
Zan ce wannan shi ne babban sakamakon gudanarwa, kuma yana haɗuwa kai tsaye da labarin ɓangare na uku. Binciken ya nuna yadda kamfanoni ke fahimtar abin da suke saye daban. Ga tsarin high-materiality, 77% na waɗanda suka amsa sun ce sun dogara da zurfin sani kan tsarin da aka haɓaka a cikin gida, idan aka kwatanta da 24% ga tsarin ɓangare na uku; ga tsarin ɓangare na uku, 81% sun dogara a maimakon haka da tabbacin mai samarwa (vendor assurance) da yarjejeniyoyin kwangila. Tsarin da aka saya maimakon a gina shi tsari ne da pre-deployment review da vendor due diligence su ne levers bayyanannu. Monitoring da recertification sun dogara ne kan ganin cikin tsari, kuma ƙarancin transparency cikin tsarin AI na ɓangare na uku 34% na waɗanda suka amsa tuni suka nuna a matsayin takura. Binciken bai faɗi wannan alaƙar dalili ba, kuma bayanan ba za su iya tabbatar da ita ba, amma guntuwar sun dace da juna ta hanyar da ba ta da daɗi.
Kamfanonin da kansu sun faɗi haka. Sa ido kan tsarin AI na ɓangare na uku shi ne fannin rashin tabbas na gudanarwa ko ƙalubalen aiwatarwa da aka fi ambata. Amfani da ƙa'idodin model risk management kan models na ɓangare na uku ko waɗanda masu samarwa suka bayar 54% na waɗanda suka amsa suka zaɓa, kuma sa ido kan tsarin da ɓangare na uku suka samar ko suka yi tasiri sosai a kansu 49%. Don haka kusan rabin kamfanoni suna ba da rahoton ƙalubale wajen amfani da ƙa'idodin model risk management kan models na masu samarwa, kuma kusan rabi wajen sa ido kan tsarin da ɓangare na uku ke samarwa ko suke yi wa tasiri sosai.
Ajiye zaruruwan ukun gefe da gefe. Ɓangare na uku suna samar da mafi yawan tsarin high-materiality. Ƙaramin rukunin masu samarwa ne ke mamaye dangantakar da ke ƙarƙashin. Kuma gudanarwar kamfanonin da kansu, mai ƙarfi a ƙofar shiga, ta fi sirara a sa ido mai ɗorewa, inda aka nuna batun ɓangare na uku a matsayin mafi wuya. Wannan ba suka ga wani kamfani ba ne; bayani ne na inda ƙoƙarin da aka bayar da rahotonsa yake idan aka kwatanta da inda fallasar da aka bayar da rahotonta take.
Abin da za a ɗauka daga ciki #
Ga shugabannin fasaha, fassarar aiki ita ce saye ya wuce gini ga tsarin da ke da muhimmanci, aƙalla a wannan samfurin: 53% na tsarin high-materiality third-party implementations ne. Don haka ƙoƙarin injiniya yana komawa kan abin da za a iya yi a kewayen tsarin da ba ka sarrafa shi ba: transparency ta kwangila, gwaji a iyaka, da monitoring da bai dogara da samun damar shiga cikin tsarin ba. Binciken bai ba da umarnin ɗaya daga cikin waɗannan ba, kuma kada a ɗauki ɗaya a matsayin sakamakon binciken. Su ne tambayoyin da alƙaluminsa ke taso.
Ga waɗanda suka kafa kamfanoni masu sayar da AI ga fannin hada-hadar kuɗi, bayanan suna nuna inda tsaiko zai kasance. Waɗanda suka amsa sun ambaci damuwa kan information security da operational resilience (40%) da ƙarancin transparency (34%) a matsayin manyan takurawa ga karɓar AI na ɓangare na uku. Mai samarwa da zai iya amsa waɗannan damuwa biyu kai tsaye yana magance abin da babban kaso na waɗanda suka amsa suka ce yana riƙe su. Kuma saboda fiye da 90% na waɗanda suka amsa sun ambaci vendor due diligence a matsayin matakin kariya na haɗarin bayanai, ya kamata mai samarwa ya sa ran za a bincike shi.
Ga duk wanda ke son karanta alƙalamin karɓuwa na 86% a matsayin alamar cewa ɓangaren mai wuya ya wuce, binciken yana ba da ra'ayi mai natsuwa. Karɓuwa ta yaɗu, amma 'yan kamfanoni kaɗan ne ke ɗaukar babban kaso na tsarin, agents galibi suna matsayin masu taimako, kuma waɗanda suka amsa suna sa ran haɗari ya tashi a kowane rukuni cikin shekara ɗaya zuwa uku masu zuwa. Waɗanda suka amsa suna kuma sa ran fa'idoji su tashi a kowane fanni, don haka babu kyakkyawan fata ko taka-tsantsan da ke da shaidar shi kaɗai.
Gargaɗin farko suna nan aiki. Wannan taƙaitaccen bayani ne na kamfanoni 196 da suka amsa, 105 kaɗai daga cikinsu suka cika cikakken binciken, kuma ba cikakken kimantawar fannin kuɗi na UK ba ne. A cikin waɗannan iyakoki, tsarin yana da daidaito har ya zama mai amfani: tsarin da kamfanoni ke ɗauka mafiya muhimmanci na ƙara zuwa daga waje, kuma sa idon da suke ba da rahotonsa ya fi rauni inda wannan dogaro ya fi girma.
Tambayar da bayanan suka taso amma ba za su iya amsawa ba ita ce abin da ke faruwa a iyakar 'yancin kai. Tare da 34% na waɗanda suka amsa ko dai suna amfani ko suna shirin amfani da agents don sadarwa da abokan ciniki ta atomatik, idan aka kwatanta da 2% kaɗai da ke tura agents masu broad autonomy a yau, binciken gaba zai nuna ko agents za su tsaya a matsayin masu taimako idan tsarin da ke bayansu, a mafi yawan lokaci, wani ne ke samar da su.
Tambayoyin da Ake Yawan Yi #
Menene binciken Bank of England da FCA kan AI a fannin hada-hadar kuɗi na UK? Shi ne bincike na huɗu na haɗin gwiwa kan artificial intelligence da machine learning a fannin hada-hadar kuɗi na UK, bayan binciken 2019, 2022 da 2024. Bugu na 2026 ya samu amsoshi 196, 105 daga cikinsu ga cikakkiyar takardar tambayoyi. Marubutan sun gabatar da shi a matsayin taƙaitaccen bayani na shaida daga kamfanonin da suka shiga, ba cikakken kimantawar fannin kuɗi na UK ba.
Nawa ne na AI na kamfanoni ke zuwa daga ɓangare na uku? Third-party implementations sun kai 39% na dukkan tsarin AI da aka bayar da rahotonsu a 2026, sama da 33% a 2024 da 17% a 2022. A cikin tsarin high-materiality kason ya kai 53%, sama da 35% a 2024.
Yaya masu samar da AI suka taru wuri ɗaya? Manyan masu samar da cloud da compute uku sun kai 79% na dangantakar cloud/compute da aka bayar da rahotonsu, kuma 98% na waɗanda suka amsa suna amfani da ɗaya daga cikinsu. Manyan masu samar da model API/hosting uku sun kai 54% na dangantakar da aka bayar da rahotonsu a wannan rukuni. Daga cikin dangantakar cloud da compute 169 da aka bayar da rahotonsu, 37% an ɗauke su masu high criticality kuma masu wuyar maye gurbi, kuma 35% kaɗai na wannan rukunin ke da exit plans da aka gwada.
Waɗanne ƙalubalen gudanarwa kamfanoni ke ba da rahotonsu game da AI na ɓangare na uku? Sa ido kan tsarin AI na ɓangare na uku shi ne fannin rashin tabbas na gudanarwa ko ƙalubalen aiwatarwa da aka fi ambata. 54% na waɗanda suka amsa sun zaɓi amfani da ƙa'idodin model risk management kan models na ɓangare na uku ko na masu samarwa, kuma 49% sun zaɓi sa ido kan tsarin da ɓangare na uku suka samar ko suka yi tasiri sosai a kansu. Ga tsarin high-materiality na ɓangare na uku, 81% sun dogara da vendor assurance da yarjejeniyoyin kwangila, yayin da 24% kaɗai suka ce suna da zurfin sani kan tsarin.
Shin AI agents suna aiki da kansu a kamfanonin kuɗi na UK? Ba safai ba. 80% na waɗanda suka amsa sun ce sun tura AI agents, kuma 80% sun ce suna da agents a production don decision support, inda agents ke ba da shawarar matakai, mutane kuma ke yanke shawara da aiwatarwa. 2% kaɗai sun ce suna da agents masu broad autonomy, kuma ko da an haɗa da tsare-tsare, bai fi 9% ba ne ke sa ran agents su amince ko su ƙi sakamakon abokan ciniki, su aiwatar da ciniki ko su sanya oda, su fara, su aiwatar ko su toshe biyan kuɗi, ko su sauya iyakoki.
Manazarta #
- Bank of England and Financial Conduct Authority, (2026). Artificial intelligence in UK financial services - 2026 ⧉. Bank of England.
- Bank of England and Financial Conduct Authority, (2024). Artificial intelligence in UK financial services - 2024 ⧉. Bank of England.
- Bank of England, Prudential Regulation Authority and Financial Conduct Authority, (2022). DP5/22 - Artificial Intelligence and Machine Learning ⧉. Bank of England.
- Bank of England, Prudential Regulation Authority and Financial Conduct Authority, (2023). FS2/23 - Artificial Intelligence and Machine Learning ⧉. Bank of England.
Bita ta ƙarshe .
Sake buga wannan labarin
Kwafa tsarin Medium
# AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau > Originally published at [https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/](https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/) Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata. Read the full article on sebastienrousseau.com: https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/
Kwafa tsarin Mastodon
AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata. https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/
Kwafa an tsara don LinkedIn
AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata. Ga abubuwan da ya kamata a lura da su na dabarun: - Menene binciken, kuma me ba shi ba. Wallafar, wato bincike na huɗu na Bank of England da FCA kan AI da machine learning a fannin hada-hadar kuɗi na UK, taƙaitaccen bayani ne na jimillar sakamakon binciken, kuma ya dace a riƙe wannan tsari a zuciya… - Karɓuwa ta kusa zama ta kowa, kuma generative AI ne ke kan gaba. Fara da kanun labarin. - Ƙalubalen materiality. Ga tarkon da aka ambata a baya. - Daga ina tsarin masu muhimmanci ke zuwa. Sakamako mafi tasiri a binciken shi ne wanda ke ginawa da gudanar da tsarin da suka fi muhimmanci. Menene hanyar ƙungiyar ku wajen magance ƙalubalen da aka kawo a wannan rubuce-rubucen? → https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/ #HaɗarinAiNaƁangareNaUku #AiGovernance #BankOfEngland #Fca #HadaHadarKuɗiNaUk Sebastien Rousseau | CC-BY-4.0
Buga wannan labari
AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau
Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata.
BibTeX
@online{rousseau2026ai,
author = {Rousseau, Sebastien},
title = {{AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau}},
year = {2026},
url = {https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/},
urldate = {2026}
}RIS
TY - GEN AU - Rousseau, Sebastien TI - AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau PY - 2026 UR - https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/ ER -
Vancouver
Rousseau S. AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau. sebastienrousseau.com. 2026 Nov 21. Available from: https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/
Chicago
Rousseau, Sebastien. "AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau." sebastienrousseau.com. November 21, 2026. https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/.
APA
Rousseau, S. (2026, November 21). AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau. sebastienrousseau.com. https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/
Sake buga wannan labari
AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau
Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata.
An lasisin wannan labari a karkashin Creative Commons Attribution 4.0 International. Sake bugawa na bukatar nuna asalin URL na asali.
AI na Ɓangare na Uku Yanzu Shi Ne Ƙalubalen Gudanarwa a Kuɗin UK — Sebastien Rousseau Ɓangare na uku na samar da 53% na tsarin AI na high-materiality a kamfanonin kuɗi na UK, kuma sa ido a kansu shi ne ƙalubalen gudanarwa da suka fi ambata. Originally published at https://sebastienrousseau.com/ha/2026-11-21-bangare-na-uku-yanzu-shi-kalubalen/ by Sebastien Rousseau. Licensed under CC-BY-4.0.
