Isazi se-Cybersecurity sihlaziya izinsongo sisebenzisa amathuluzi e-AI akhiqizayo.

Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity?

Isingeniso

I-Generative AI - izinhlelo zobuhlakani bokwenziwa ezikwazi ukudala okuqukethwe okusha noma izibikezelo - ivela njengamandla aguqulayo ku-cybersecurity. Amathuluzi afana ne-OpenAI's GPT-4 abonise amandla okuhlaziya idatha eyinkimbinkimbi futhi akhiqize umbhalo ofana nomuntu, okuvumela izindlela ezintsha zokuzivikela ezinsongweni ze-cyber. Ochwepheshe be-Cybersecurity kanye nabathatha izinqumo zebhizinisi kuzo zonke izimboni bahlola ukuthi i-AI ekhiqizayo ingaqinisa kanjani ukuzivikela ekuhlaselweni okuvelayo. Kusukela kwezezimali nokunakekelwa kwezempilo kuya ekudayiseni nakuhulumeni, izinhlangano kuyo yonke imikhakha zibhekana nemizamo yobugebengu bokweba imininingwane ebucayi, uhlelo olungayilungele ikhompuyutha, nezinye izinsongo ezingase zisize ukulwa ne-AI. Kuleli phepha elimhlophe, sihlola ukuthi i-AI ekhiqizayo ingasetshenziswa kanjani ku-cybersecurity , sigqamisa izinhlelo zokusebenza zomhlaba wangempela, okungenzeka esikhathini esizayo, nokucatshangelwa okubalulekile kokutholwa.

I-Generative AI ihlukile ku-AI yokuhlaziya evamile ngokuthola amaphethini kuphela kodwa nokudala okuqukethwe - noma ngabe ukulingisa ukuhlasela ukuze kuqeqeshe ukuzivikela noma ukukhiqiza izincazelo zolimi lwemvelo zedatha yezokuphepha eyinkimbinkimbi. Leli khono elikabili liyenza inkemba esika nhlangothi zombili: linikeza amathuluzi amasha anamandla okuzivikela, kodwa abalingisi abasabisayo bangawasebenzisa futhi. Izigaba ezilandelayo zihlola ububanzi obubanzi bamacala okusetshenziswa kwe-AI ekhiqizayo ku-cybersecurity, kusukela ekutholeni ubugebengu bokweba imininingwane ebucayi ngokuzenzakalelayo kuya ekuthuthukiseni impendulo yesigameko. Siphinde sixoxe ngezinzuzo lezi zinto eziqanjiwe ze-AI ezithembisayo, ezihambisana nezingozi (njengokuthi i-AI “imibono engemihle” noma ukusetshenziswa kabi kwezitha) okumele izinhlangano zilawule. Okokugcina, sinikeza okuthathayo okusebenzayo ukusiza amabhizinisi ukuthi ahlole futhi ahlanganise ngokunesibopho i-AI ekhiqizayo emasu awo okuphepha ku-inthanethi.

I-Generative AI ku-Cybersecurity: Uhlolojikelele

I-Generative AI ku-cybersecurity ibhekisela kumamodeli e-AI - ngokuvamile amamodeli amakhulu olimi noma amanye amanethiwekhi we-neural - angakhiqiza imininingwane, izincomo, ikhodi, noma idatha yokwenziwa yokusiza emisebenzini yezokuphepha. Ngokungafani namamodeli aqagelayo kuphela, i-AI ekhiqizayo ingalingisa izimo futhi ikhiqize imiphumela engafundeka umuntu (isb. imibiko, izexwayiso, noma amasampula ekhodi anonya) ngokusekelwe kudatha yayo yokuqeqeshwa. Leli khono lisetshenziswa ukuze libikezele, libone, futhi liphendule ezinsongweni ngezindlela eziguquguqukayo kakhulu kunangaphambili ( Iyini I-Generative AI ku-Cybersecurity? - Palo Alto Networks ). Isibonelo, amamodeli akhiqizayo angahlaziya amalogi amakhulu noma izinqolobane zobuhlakani ezisongela futhi akhiqize isifinyezo esifushane noma isenzo esinconyiwe, asebenze cishe “njengomsizi” we-AI emaqenjini okuvikela.

Ukuqaliswa kwangaphambi kwesikhathi kwe-AI ekhiqizayo yokuzivikela ku-inthanethi kubonise isithembiso. Ngo-2023, iMicrosoft yethula i-Security Copilot , umsizi onikwe amandla we-GPT-4 wabahlaziyi bezokuphepha, ukusiza ukukhomba ukwephulwa nokuhlunga amasiginali we-Microsoft angama-trillion angama-65 nsuku zonke ( I-Microsoft Security Copilot ingumsizi omusha we-GPT-4 AI we-cybersecurity | The Verge ). Abahlaziyi bangakwazi ukwazisa le sistimu ngolimi lwemvelo (isb. “Fingqa zonke izigameko zokuphepha emahoreni angu-24 adlule” ), futhi umshayeli uzokhipha isifinyezo esiwusizo sokulandisa. Ngokufanayo, i-Google's Threat Intelligence AI isebenzisa imodeli ekhiqizayo ebizwa ngokuthi i-Gemini ukuze unike amandla ukusesha kwengxoxo ngokusebenzisa isizindalwazi se-Google esisongelayo esikhulu, ihlaziye ngokushesha ikhodi esolisayo futhi ifinyeze okutholakele ukusiza abazingeli be-malware ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Eziyi-10 Zangempela Zomhlaba ). Lezi zibonelo zibonisa amandla: i-AI ekhiqizayo ingagaya idatha eyinkimbinkimbi, enkulu yokuphepha ku-inthanethi futhi yethule imininingwane ngendlela efinyelelekayo, isheshise ukwenza izinqumo.

Ngesikhathi esifanayo, i-AI ekhiqizayo ingakha okuqukethwe okungelona iqiniso okungokoqobo, okuyinzuzo yokulingisa nokuqeqeshwa (futhi, ngeshwa, kubahlaseli abakha ubunjiniyela bomphakathi). Njengoba siqhubekela ezimweni ezithile zokusetshenziswa, sizobona ukuthi ikhono le-AI elikhiqizayo lokuhlanganisa nokuhlaziya ulwazi lisekela izinhlelo zayo eziningi zokusebenza zokuphepha ku-inthanethi . Ngezansi, singena ezimweni zokusetshenziswa ezibalulekile, sihlanganisa yonke into kusukela ekuvimbeleni ubugebengu bokweba imininingwane ebucayi kuya ekuthuthukisweni kwesofthiwe evikelekile, ngezibonelo zokuthi ngayinye isetshenziswa kanjani kuzo zonke izimboni.

Izicelo Ezibalulekile Ze-Generative AI ku-Cybersecurity

Umfanekiso: Izimo eziyinhloko zokusetshenziswa kwe-AI ekhiqizayo ku-cybersecurity zihlanganisa amakhophi e-AI emaqenjini okuvikela, ukuhlaziya ubungozi bekhodi, ukutholwa kokusongelwa okuguquguqukayo, ukulingisa ukuhlasela kosuku oluyiziro, ukuphepha okuthuthukisiwe kwe-biometric, nokutholwa kobugebengu bokweba imininingwane ebucayi ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ).

Ukutholwa Nokuvimbela Ubugebengu Bokweba imininingwane ebucayi

Ubugebengu bokweba imininingwane ebucayi busengenye yezinsongo ze-inthanethi ezigcwele kakhulu, ezikhohlisa abasebenzisi ukuthi bachofoze izixhumanisi ezinonya noma badalule imininingwane. I-Generative AI isatshalaliswa kukho kokubili imizamo yobugebengu bokweba imininingwane ebucayi futhi iqinise ukuqeqeshwa kwabasebenzisi ukuze kuvinjelwe ukuhlasela okuyimpumelelo. Ngasohlangothini lokuzivikela, amamodeli e-AI angahlaziya okuqukethwe kwe-imeyili kanye nokuziphatha komthumeli ukuze abone izimpawu ezicashile zobugebengu bokweba imininingwane ebucayi lezo izihlungi ezisekelwe emthethweni ezingase ziphuthelwe. Ngokufunda kumasethi edatha amakhulu wama-imeyili asemthethweni aphikisana nomgunyathi, imodeli ekhiqizayo ingahlaba umkhosi okudidayo ngephimbo, amagama, noma umongo obonisa umkhonyovu - ngisho noma uhlelo lolimi nesipelingi lungasakuniki. Eqinisweni, abacwaningi be-Palo Alto Networks bayaqaphela ukuthi i-AI ekhiqizayo ingakwazi ukukhomba "izimpawu ezicashile zama-imeyili obugebengu bokweba imininingwane ebucayi okungenzeka zingabonakali," izinhlangano ezisiza izinhlangano zihlale isinyathelo esisodwa ngaphambi kwabakhwabanisi ( Iyini I-Generative AI ku-Cybersecurity? - I-Palo Alto Networks ).

Amaqembu ezokuphepha asebenzisa ne-generative AI ukuze alingise ukuhlaselwa kobugebengu bokweba imininingwane ebucayi ukuze aqeqeshelwe futhi ahlaziywe. Isibonelo, i-Ironscales yethule ithuluzi lokulingisa ubugebengu bokweba imininingwane ebucayi elinikwe amandla e-GPT elikhiqiza ngokuzenzakalelayo ama-imeyili omgunyathi aklanyelwe abasebenzi benhlangano ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Lawa ma-imeyili enziwe nge-AI abonisa amaqhinga omhlaseli akamuva, anikeza abasebenzi ukuzijwayeza okungokoqobo ekuboneni okuqukethwe kobugebengu. Ukuqeqeshwa okunjalo okuqondene nawe kubalulekile njengoba abahlaseli ngokwabo besebenzisa i-AI ukuze bakhe imihuzuko ekholisayo. Ngokuphawulekayo, ngenkathi i-AI ekhiqizayo ingakhiqiza imilayezo yobugebengu bokweba imininingwane ebucayi (sezihambile izinsuku zesiNgisi esiphuka kalula), abavikeli bathole ukuthi i-AI ayihluleki. Ngo-2024, abacwaningi bezokuphepha be-IBM benza uhlolo oluqhathanisa ama-imeyili obugebengu bokweba imininingwane ebucayi abhalwe abantu nalawo akhiqizwe yi-AI, futhi “okumangalisayo ukuthi ama-imeyili akhiqizwe nge-AI ayesawathola kalula naphezu kolimi lwawo olulungile” ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Lokhu kuphakamisa ukuthi ukwazi komuntu kuhlanganiswe nokutholwa okusizwa yi-AI kusengakwazi ukubona ukungqubuzana okucashile noma amasiginali wemethadatha emikhonyongweni ebhalwe nge-AI.

I-Generative AI isiza ukuvikela ubugebengu bokweba imininingwane ebucayi ngezinye izindlela, futhi. Amamodeli angasetshenziswa ukwenza izimpendulo ezizenzakalelayo noma izihlungi ezihlola ama-imeyili asolisayo. Isibonelo, isistimu ye-AI ingaphendula i-imeyili ngemibuzo ethile ukuze iqinisekise ukufaneleka komthumeli noma isebenzise i-LLM ukuze ihlaziye izixhumanisi ze-imeyili nezinanyathiselwa ku-sandbox, bese ifingqa noma iyiphi inhloso enonya. Inkundla yezokuphepha ye-NVIDIA i-Morpheus ibonisa amandla e-AI kule nkundla – isebenzisa amamodeli e-NLP akhiqizayo ukuze ihlaziye ngokushesha futhi ihlukanise ama-imeyili, futhi itholwe ithuthukisa ukutholwa kwe-imeyili yobugebengu bokweba imininingwane ebucayi ngo- 21% uma iqhathaniswa namathuluzi okuvikela endabuko ( 6 Sebenzisa Amakesi E-Generative AI ku-Cybersecurity [+ Izibonelo] ). U-Morpheus uveza ngisho namaphethini okuxhumana nabasebenzisi ukuze athole ukuziphatha okungajwayelekile (njengomsebenzisi othumela i-imeyili ngokuzumayo amakheli amaningi angaphandle), okungabonisa i-akhawunti eyonakele ethumela ama-imeyili obugebengu bokweba imininingwane ebucayi.

Empeleni, izinkampani ezimbonini zonkana seziqala ukwethemba i-AI ukuthi izohlunga i-imeyili ne-web traffic ukuze kuhlaselwe ubunjiniyela bezenhlalo. Amafemu ezezimali, isibonelo, asebenzisa i-AI ekhiqizayo ukuskena ukuxhumana ukuze kutholakale imizamo yokuzenza ongeyena engaholela ekukhwabaniseni ngocingo, kuyilapho abahlinzeki bezempilo besebenzisa i-AI ukuze bavikele idatha yesiguli ekuphuleni umthetho okuhlobene nobugebengu bokweba imininingwane ebucayi. Ngokukhiqiza izimo ezingokoqobo zobugebengu bokweba imininingwane ebucayi kanye nokuhlonza izimpawu zemilayezo enonya, i-AI ekhiqizayo yengeza isendlalelo esinamandla kumasu okuvimbela ubugebengu bokweba imininingwane ebucayi. I-takeaway: I-AI ingasiza ekutholeni nasekukhipheni izikhali ukuhlaselwa kobugebengu bokweba imininingwane ebucayi ngokushesha nangokunembe kakhudlwana, njengoba abahlaseli besebenzisa ubuchwepheshe obufanayo ukuthuthukisa igeyimu yabo.

Ukutholwa kohlelo olungayilungele ikhompyutha nokuhlaziywa kosongo

Uhlelo olungayilungele ikhompuyutha lwesimanje luhlala luvela - abahlaseli bakhiqiza okuhlukile okusha noma ikhodi efiphazayo ukuze badlule amasignesha okuvikela amagciwane. I-Generative AI inikeza amasu anoveli akho kokubili ukuthola uhlelo olungayilungele ikhompuyutha nokuqonda ukuziphatha kwayo. Enye indlela ukusebenzisa i-AI ukwenza “amawele amabi” ohlelo olungayilungele ikhompuyutha : abacwaningi bezokuvikela bangaphakela isampula eyaziwayo yohlelo olungayilungele ikhompuyutha benze imodeli ekhiqizayo ukuze bakhe izinhlobo eziningi eziguquliwe zalolo hlelo olungayilungele ikhompuyutha. Ngokwenza kanjalo, balindele ngempumelelo ama-tweaks umhlaseli angase awenze. Lokhu okuhlukile okukhiqizwa yi-AI kungase kusetshenziselwe ukuqeqesha isivikeli magciwane kanye nezinhlelo zokutholwa kokungena, ukuze ngisho nezinguqulo eziguquliwe zohlelo olungayilungele ikhompuyutha zibonakale endle ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Leli su elisebenzayo lisiza ukunqamula umjikelezo lapho izigebengu ze-inthanethi zishintsha kancane uhlelo olungayilungele ikhompuyutha ukuze zigweme ukutholwa futhi abavikeli kufanele baxabane ukuze babhale amasiginesha amasha isikhathi ngasinye. Njengoba kuphawulwe kwenye i-podcast yomkhakha, ochwepheshe bezokuphepha manje basebenzisa i-AI ekhiqizayo ukuze "balingise ithrafikhi yenethiwekhi futhi bakhiqize imithwalo ekhokhelwayo enonya elingisa ukuhlasela okuyinkimbinkimbi," okuhlola ingcindezi yokuzivikela ngokumelene nomndeni wonke wezinsongo esikhundleni sesenzakalo esisodwa. Lokhu kutholwa kokusongelwa okuguquguqukayo kusho ukuthi amathuluzi okuphepha azomelana kakhulu ne-polymorphic malware ebingase ingene.

Ngaphandle kokutholwa, i-AI ekhiqizayo isiza ekuhlaziyeni uhlelo olungayilungele ikhompuyutha kanye nobunjiniyela obuhlehlisayo , ngokuvamile okuyimisebenzi edinga abasebenzi abaningi yabahlaziyi bosongo. Izibonelo zolimi ezinkulu zinganikezwa umsebenzi wokuhlola amakhodi noma imibhalo esolisayo futhi ichaze ngolimi olulula ukuthi ikhodi ihloselwe ukwenzani. Isibonelo somhlaba wangempela i- VirusTotal Code Insight , isici se-VirusTotal ye-Google esebenzisa imodeli ekhiqizayo ye-AI (i-Google's Sec-PaLM) ukuze kukhiqizwe izifinyezo zolimi lwemvelo zekhodi engase ibe yingozi ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Empeleni “iwuhlobo lwe-ChatGPT olunikezelwe ekubhalweni kwezokuphepha,” esebenza njengomhlaziyi we-malware we-AI osebenza 24/7 ukuze asize abahlaziyi abangabantu baqonde izinsongo ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Esikhundleni sokubuka umbhalo ongajwayelekile noma ikhodi kanambambili, ilungu lethimba lezokuphepha lingathola incazelo ngokushesha ku-AI - isibonelo, "Lo mbhalo uzama ukulanda ifayela kusuka kuseva ye-XYZ bese ulungisa izilungiselelo zesistimu, okuyinkomba yokuziphatha kwe-malware." Lokhu kusheshisa kakhulu impendulo yesigameko, njengoba abahlaziyi bengakwazi ukuhlola futhi baqonde uhlelo olungayilungele ikhompuyutha entsha ngokushesha kunangaphambili.

I-Generative AI iphinde isetshenziselwe ukukhomba uhlelo olungayilungele ikhompuyutha kumadathasethi amakhulu . Izinjini zokuvikela amagciwane ezivamile ziskena amafayela ukuze zithole amasiginesha aziwayo, kodwa imodeli ekhiqizayo ingahlola izici zefayela futhi ibikezele ukuthi ingabe inonya ngokusekelwe kumaphethini afundiwe. Ngokuhlaziya izici zezigidigidi zamafayela (anonya futhi ayingozi), i-AI ingase ibambe inhloso enonya lapho kungekho siginesha ecacile ekhona. Isibonelo, imodeli ekhiqizayo ingahlaba umkhosi okusebenzisekayo njengokusolisayo ngenxa yokuthi iphrofayela yayo yokuziphatha "ibukeka" njengokuhluka okuncane kwe-ransomware eyakubona ngesikhathi sokuqeqeshwa, nakuba kanambambili kukusha. Lokhu kutholwa okusekelwe ekuziphatheni kusiza ukuphikisa inoveli noma uhlelo olungayilungele ikhompuyutha losuku oluyiziro. I-Threat Intelligence AI ye-Google (ingxenye ye-Chronicle/Mandiant) kubikwa ukuthi isebenzisa imodeli yayo ekhiqizayo ukuze ihlaziye ikhodi engase ibe yingozi futhi “isize ochwepheshe bezokuphepha ngokuphumelela kakhudlwana nangempumelelo ekulweni nohlelo olungayilungele ikhompuyutha kanye nezinye izinhlobo zezinsongo.” ( Ingasetshenziswa Kanjani I-Generative AI ku-Cybersecurity? Izibonelo ze-Real-World 10 ).

Ngakolunye uhlangothi, kufanele sivume ukuthi abahlaseli bangasebenzisa i-AI ekhiqizayo nalapha - ukuze bakhe ngokuzenzakalelayo uhlelo olungayilungele ikhompuyutha oluzijwayeza lona. Eqinisweni, ochwepheshe bezokuphepha baxwayisa ngokuthi i-AI ekhiqizayo ingasiza izigebengu ze-inthanethi ukuthi zithuthukise uhlelo olungayilungele ikhompuyutha okunzima ukulithola ( Iyini I-Generative AI ku-Cybersecurity? - Palo Alto Networks ). Imodeli ye-AI ingayalwa ukuthi iguqule ucezu lohlelo olungayilungele ikhompuyutha ngokuphindaphindiwe (ukushintsha ukwakheka kwefayela layo, izindlela zokubethela, njll.) ize igweme konke ukuhlolwa okwaziwayo kokuvikela amagciwane. Lokhu kusetshenziswa okuphikisayo kuwukukhathazeka okukhulayo (ngezinye izikhathi okubizwa ngokuthi “i-AI-powered malware” noma i-polymorphic malware njengesevisi). Sizoxoxa ngezingozi ezinjalo kamuva, kodwa igcizelela ukuthi i-AI ekhiqizayo iyithuluzi kulo mdlalo wekati negundane osetshenziswa yibo bobabili abavikeli nabahlaseli.

Sekukonke, i-AI ekhiqizayo ithuthukisa ukuvikela uhlelo olungayilungele ikhompuyutha ngokunika amandla amaqembu okuvikela ukuthi acabange njengomhlaseli - ekhiqiza izinsongo ezintsha nezisombululo endlini. Ukuthi ikhiqiza uhlelo olungayilungele ikhompuyutha yokwenziwa ukuze kuthuthukiswe izinga lokutholwa noma isebenzisa i-AI ukuchaza futhi iqukethe uhlelo olungayilungele ikhompuyutha lwangempela olutholakala kumanethiwekhi, lawa maqhinga asebenza kuzo zonke izimboni. Ibhange lingase lisebenzise ukuhlaziywa kohlelo olungayilungele ikhompuyutha olushayelwa yi-AI ukuze lihlaziye ngokushesha i-macro esolisayo kusipredishithi, kuyilapho inkampani ekhiqizayo ingase ithembele ku-AI ukuze ithole uhlelo olungayilungele ikhompuyutha oluqondise kumasistimu okulawula izimboni. Ngokwandisa ukuhlaziya okujwayelekile kohlelo olungayilungele ikhompuyutha nge-AI ekhiqizayo, izinhlangano zingaphendula imikhankaso yohlelo olungayilungele ikhompuyutha ngokushesha nangokuqhubekayo kunangaphambili.

I-Treat Intelligence kanye nokuhlaziya okuzenzakalelayo

Nsuku zonke, izinhlangano zihlushwa idatha yezobunhloli ezisongelayo - kusukela kokuphakelayo kwezinkomba ezisanda kutholwa zokuvumelana (i-IOCs) kuya kumibiko yomhlaziyi mayelana namaqhinga ezigebengu ezisafufusa. Inselele yamaqembu ezokuphepha ihlunga lolu lwazi oluningi kanye nokukhipha imininingwane engenzeka. I-Generative AI ibonakala ibalulekile ekuziqhubeleni phambili ekuhlaziyeni nasekusetshenzisweni kobuhlakani obusongelayo . Esikhundleni sokufunda mathupha inqwaba yemibiko noma okufakiwe kusizindalwazi, abahlaziyi bangasebenzisa i-AI ukuze bafingqe futhi bahlanganise intel yosongo ngesivinini somshini.

Isibonelo esisodwa esiphathekayo i-Threat Intelligence suite ye-Google, ehlanganisa i-AI ekhiqizayo (imodeli ye-Gemini) nedatha ye-Google yokusongela evela ku-Mandiant ne-VirusTotal. Le AI ihlinzeka “ngosesho lwengxoxo kuyo yonke inqolobane enkulu ye-Google yobuhlakani obusongelayo” , evumela abasebenzisi ukuthi babuze imibuzo yemvelo mayelana nezinsongo futhi bathole izimpendulo ezicwebile ( I-Generative AI Ingasetshenziswa Kanjani Ku-Cybersecurity? Izibonelo Zomhlaba Wangempela eziyi-10 ). Isibonelo, umhlaziyi angabuza, "Siyibonile noma iyiphi i-malware ehlobene ne-Threat Group X eqondise imboni yethu?" futhi i-AI izodonsa i-intel efanele, mhlawumbe iphawule ukuthi “Yebo, Iqembu X Losongo lixhunywe kumkhankaso wobugebengu bokweba imininingwane ebucayi ngenyanga edlule kusetshenziswa i-malware Y” , kanye nesifinyezo sokuziphatha kwalolo hlelo olungayilungele ikhompuyutha. Lokhu kunciphisa kakhulu isikhathi sokuqoqa imininingwane engadinga ukubuza amathuluzi amaningi noma ukufunda imibiko emide.

I-Generative AI ingaphinda ihlobanise futhi ifinyeze amathrendi okusongela . Ingase ihlanganise izinkulungwane zokuthunyelwe kwebhulogi yezokuphepha, izindaba eziphula umthetho, nengxoxo emnyama yewebhu bese ikhiqiza isifinyezo esiphezulu “sezinsongo eziphezulu ze-cyber kuleli sonto” esithangamini se-CISO. Ngokwesiko, leli zinga lokuhlaziya nokubika lithathe umzamo omkhulu womuntu; manje imodeli ecushwe kahle ingakwazi ukuyibhala ngemizuzwana, abantu becwenga okukhiphayo kuphela. Izinkampani ezifana ne-ZeroFox zenze i-FoxGPT , ithuluzi elikhiqizayo le-AI eliklanyelwe ngokukhethekile “ukusheshisa ukuhlaziywa nokufingqwa kobuhlakani kuwo wonke amasethi edatha amakhulu,” okuhlanganisa okuqukethwe okunonya kanye nedatha yobugebengu bokweba imininingwane ebucayi ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? 10 Izibonelo Zomhlaba Wangempela ). Ngokwenza ngokuzenzakalela ukuphakamisa okusindayo kwedatha yokufunda nereferensi, i-AI inika amandla amaqembu e-intel asongelayo ukuthi agxile ekwenzeni izinqumo nasekuphenduleni.

Esinye isimo esisetshenziswayo ukuzingela usongo lwengxoxo . Cabanga ngomhlaziyi wezokuphepha esebenzisana nomsizi we-AI: “Ngibonise noma yiziphi izimpawu zokukhishwa kwedatha emahoreni angu-48 adlule” noma “Iziphi izihlaseli ezintsha eziphezulu zobungozi abazisebenzisayo kuleli sonto?” I-AI ingahumusha umbuzo, iseshe izingodo zangaphakathi noma imithombo ye-intel yangaphandle, futhi iphendule ngempendulo ecacile noma ngisho nohlu lwezigameko ezifanele. Lokhu akwenzeki kude – izinhlelo zesimanjemanje zolwazi lwezokuphepha kanye nokuphathwa kwemicimbi (SIEM) seziqala ukuhlanganisa imibuzo yolimi lwemvelo. I-IBM's QRadar security suite, ngokwesibonelo, yengeza izici ezikhiqizayo ze-AI ngo-2024 ukuze ivumele abahlaziyi “babuze […] imibuzo ethize mayelana nendlela yokuhlasela efingqiwe” yesigameko futhi bathole izimpendulo ezinemininingwane. Ingase futhi "ihumushe futhi ifinyeze ubuhlakani obusongelayo obuhlobene kakhulu" ngokuzenzakalelayo ( Ingasetshenziswa Kanjani I-AI Yokukhiqiza Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Empeleni, i-AI ekhiqizayo ishintsha izintaba zedatha yobuchwepheshe ibe imininingwane yosayizi wengxoxo lapho kudingeka khona.

Kuzo zonke izimboni, lokhu kunemiphumela emikhulu. Umhlinzeki wokunakekelwa kwezempilo angasebenzisa i-AI ukuze ahlale enolwazi lwakamuva emaqenjini e-ransomware aqondise izibhedlela, ngaphandle kokunikeza umhlaziyi ocwaningweni lwesikhathi esigcwele. I-SOC yenkampani edayisayo ingafingqa ngokushesha amaqhinga amasha e-POS yohlelo olungayilungele ikhompuyutha lapho itshela abasebenzi basesitolo se-IT. Futhi kuhulumeni, lapho idatha yokusongela evela kuma-ejensi ahlukahlukene kufanele ihlanganiswe, i-AI ingakhiqiza imibiko ebumbene egqamisa izexwayiso ezibalulekile. Ngokuzenzakalela ukuqoqwa kobuhlakani obusongelayo kanye nencazelo , i-AI ekhiqizayo isiza izinhlangano ukuthi zisabele ngokushesha ezinsongweni ezivelayo futhi inciphise ubungozi bokuphuthelwa izexwayiso ezibalulekile ezifihlwe emsindweni.

Ukuthuthukiswa Kwesikhungo Sezokuvikela (SOC).

Izikhungo Zokusebenza Kwezokuphepha zidume kabi ngokukhathala kwesixwayiso kanye nenani elicindezelayo ledatha. Umhlaziyi we-SOC ojwayelekile angase adlule ezinkulungwaneni zezaziso nezehlakalo usuku ngalunye, ephenya izehlakalo ezingaba khona. I-Generative AI isebenza njengesiphindaphinda samandla kuma-SOC ngokuzenzela umsebenzi ojwayelekile, inikeze izifinyezo ezihlakaniphile, ngisho nokuhlela ezinye izimpendulo. Umgomo uwukuthuthukisa ukugeleza komsebenzi we-SOC ukuze abahlaziyi abangabantu bakwazi ukugxila ezindabeni ezibucayi kakhulu kuyilapho umsizi we-AI ephatha okunye.

Uhlelo lokusebenza olulodwa olukhulu lusebenzisa i-AI ekhiqizayo “njengeCopilot Yomhlaziyi” . I-Microsoft's Security Copilot, ephawulwe ngaphambili, iyisibonelo salokhu: “iklanyelwe ukusiza umsebenzi womhlaziyi wezokuphepha esikhundleni sokuyishintsha,” isiza ngophenyo lwezigameko nokubika ( Microsoft Security Copilot ingumsizi omusha we-GPT-4 AI we-cybersecurity | The Verge ). Empeleni, lokhu kusho ukuthi umhlaziyi angafaka idatha eluhlaza - amalogi okuvikela umlilo, umugqa wesikhathi womcimbi, noma incazelo yesigameko - futhi acele i-AI ukuthi iyihlaziye noma iyifingqe. Umshayeli angase akhiphe ukulandisa okufana nokuthi, "Kubonakala sengathi ngo-2:35 AM, ukungena ngemvume okusolisayo okuvela ku-IP X kuphumelele kuseva Y, kulandelwa ukudluliswa kwedatha okungajwayelekile, okubonisa ukwephulwa okungenzeka kwaleyo seva." Lolu hlobo lokulungisa isimo ngokushesha lubaluleke kakhulu uma isikhathi siwumongo.

Abashayeli bezindiza be-AI futhi basiza ukwehlisa umthwalo we-triage 1. Ngokusho kwedatha yomkhakha, ithimba lezokuphepha lingachitha amahora angu-15 ngeviki livele lihlunga izexwayiso ezingaba ngu-22,000 kanye nokuhle okungamanga ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Nge-generative AI, eziningi zalezi zixwayiso zingacutshungulwa ngokuzenzakalelayo - i-AI ingachitha lezo ezinobuntu ngokusobala (ngokucabanga okunikeziwe) futhi igqamise lezo ezidinga ukunakwa ngempela, kwesinye isikhathi ize iphakamise nokubalulekile. Eqinisweni, amandla e-AI okukhiqizayo ekuqondeni umongo asho ukuthi ingahlobanisa izexwayiso ezingase zibonakale zingenabungozi uma zodwa kodwa zihlangene zikhombise ukuhlasela kwezigaba eziningi. Lokhu kunciphisa ithuba lokuphuthelwa ukuhlaselwa ngenxa “yokukhathala okuqwashile.”

Abahlaziyi be-SOC nabo basebenzisa ulimi lwemvelo ne-AI ukusheshisa ukuzingela nophenyo. -SentinelOne's Purple AI , ngokwesibonelo, ihlanganisa ukusebenzelana okusekelwe ku-LLM nedatha yesikhathi sangempela yokuphepha, okuvumela abahlaziyi ukuthi "babuze imibuzo eyinkimbinkimbi yokuzingela izinsongo ngesiNgisi esilula futhi bathole izimpendulo ezisheshayo, ezinembile" ( I-Generative AI Ingasetshenziswa Kanjani Ku-Cybersecurity? Izibonelo Zomhlaba Wangempela eziyi-10 ). Umhlaziyi angathayipha, “Ingabe kukhona amaphoyinti okugcina axhunywe nesizinda badguy123[.]com enyangeni edlule?” , kanye ne-Purple AI izosesha amalogi ukuze iphendule. Lokhu kusindisa umhlaziyi ekubhaleni imibuzo yesizindalwazi noma imibhalo - i-AI ikwenza ngaphansi kwe-hood. Kuphinde kusho ukuthi abahlaziyi abancane bangakwazi ukusingatha imisebenzi ebidinga unjiniyela ongumakadebona onekhono lokubuza imibuzo, bakhuphule amakhono eqembu ngempumelelo ngosizo lwe-AI . Ngempela, abahlaziyi babika ukuthi isiqondiso esikhiqizayo se-AI "sithuthukisa amakhono abo kanye nekhono labo" , njengoba abasebenzi abancane manje bangathola ukwesekwa kwekhodi okufunwayo noma amathiphu okuhlaziya ku-AI, ukunciphisa ukuthembela ekuceleni njalo amalungu eqembu eliphezulu usizo ( 6 Sebenzisa Amacala E-Generative AI ku-Cybersecurity [+ Izibonelo] ).

Okunye ukulungiselelwa kwe-SOC ukufinyezwa kwesigameko okuzenzakalelayo kanye nemibhalo . Ngemva kokusingathwa kwesigameko, othile kufanele abhale umbiko - umsebenzi abaningi abawuthola uyisicefe. I-Generative AI ingathatha idatha ye-forensic (amalogi esistimu, ukuhlaziya uhlelo olungayilungele ikhompuyutha, umugqa wesikhathi wezenzo) futhi ikhiqize umbiko wesigameko osalungiswa. I-IBM yakha leli khono libe yi-QRadar ukuze “ngokuchofoza okukodwa” isifinyezo sesigameko sikhiqizelwe ababambiqhaza abahlukene (abaphathi, amaqembu e-IT, njll.) ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Lokhu akugcini nje kuphela kongi isikhathi kodwa futhi kuqinisekisa ukuthi akukho okushaywa indiva embikweni, njengoba i-AI ingafaka yonke imininingwane efanele ngokungaguquki. Ngokunjalo, ukuze kuthotshelwe futhi kucwaningwe, i-AI ingagcwalisa amafomu noma amathebula obufakazi ngokusekelwe kudatha yesigameko.

Imiphumela yomhlaba wangempela iyaphoqa. Abamukela kuqala i-SOAR ye-Swimlane eqhutshwa yi-AI (i-orchestration yezokuvikela, i-automation, kanye nempendulo) babika izinzuzo ezinkulu zokukhiqiza - i-Global Data Systems, isibonelo, ibone ithimba labo le-SecOps liphatha umthwalo wamacala amakhulu kakhulu; omunye umqondisi uthe "engikwenzayo namuhla nabahlaziyi be-7 cishe kuzothatha abasebenzi be-20 ngaphandle" kwe-AI-powered automation ( Ingasetshenziswa Kanjani I-Generative AI ku-Cybersecurity ). Ngamanye amazwi, i-AI ku-SOC ingaphindaphinda umthamo . Kuzo zonke izimboni, kungakhathaliseki ukuthi inkampani yezobuchwepheshe esebenza ngezexwayiso zokuphepha kwamafu noma imboni egada amasistimu e-OT, amaqembu e-SOC ame ukuze athole ukutholwa nokusabela ngokushesha, izigameko ezimbalwa ezigejiwe, kanye nemisebenzi esebenza kahle kakhulu ngokwamukela abasizi be-AI abakhiqizayo. Imayelana nokusebenza ngobuhlakani - ukuvumela imishini ukuthi isingathe imisebenzi ephindaphindwayo nesindayo yedatha ukuze abantu bakwazi ukusebenzisa ulwazi lwabo nolwazi lapho kubaluleke kakhulu.

Ukuphathwa Kobungozi kanye Nokulingiswa Kosongo

Ukuhlonza nokuphatha ubungozi - ubuthakathaka kusofthiwe noma izinhlelo abahlaseli abangazisebenzisa - kuwumsebenzi oyinhloko wokuphepha ku-inthanethi. I-Generative AI ithuthukisa ukuphathwa kobungozi ngokusheshisa ukutholwa, isiza ekubekeni phambili iziqephu, futhi ilingise nokuhlasela kulokho kuba sengozini ukuze kuthuthukiswe ukulungela. Empeleni, i-AI isiza izinhlangano ukuthi zithole futhi zilungise izimbobo zezikhali zazo ngokushesha okukhulu, futhi zihlole ukuzivikela ngaphambi kokuba abahlaseli bangempela benze kanjalo.

Olunye uhlelo lokusebenza olubalulekile lusebenzisa i-AI ekhiqizayo ukuze kubuyekezwe ikhodi ngokuzenzakalelayo kanye nokutholwa kokuba sengozini . Izisekelo zekhodi ezinkulu (ikakhulukazi izinhlelo zefa) zivame ukuba namaphutha okuphepha angabonakali. Amamodeli e-AI akhiqizayo angaqeqeshelwa izinqubo zokubhala ezivikelekile namaphethini eziphazamisi ezivamile, bese zidedelwa ngekhodi yomthombo noma okuhlanganisiwe okumbambambili ukuze kutholwe ubungozi obungaba khona. Isibonelo, abacwaningi be-NVIDIA bathuthukise ipayipi le-AI elikhiqizayo elingahlaziya iziqukathi zesofthiwe yefa futhi likhombe ubungozi “ngokunemba okuphezulu - kufika ku-4× ngokushesha kunochwepheshe abangabantu.” ( 6 Sebenzisa Amakesi E-Generative AI ku-Cybersecurity [+ Izibonelo] ). I-AI empeleni yafunda ukuthi ikhodi engavikelekile ibukeka kanjani futhi yakwazi ukuskena ngesofthiwe yamashumi eminyaka ubudala ukuze ihlabe umkhosi imisebenzi eyingozi nemitapo yolwazi, isheshisa kakhulu inqubo evamile ehamba kancane yokuhlolwa kwekhodi ngesandla. Lolu hlobo lwethuluzi lungashintsha umdlalo ezimbonini ezifana nezezimali noma zikahulumeni ezithembele kumakhodi amakhulu, amadala - i-AI isiza ukwenza ukuphepha kube okwakamuva ngokumba izinkinga abasebenzi abangathatha izinyanga noma iminyaka ukuze bathole (uma kwenzeka).

I-Generative AI iphinde isize ekulawuleni ubungozi bokugeleza komsebenzi ngokucubungula imiphumela yokuskena yokuba sengozini nokuyibeka phambili. Amathuluzi afana ne-Tenable's ExposureAI asebenzisa i-AI ekhiqizayo ukuze avumele abahlaziyi babuze ngedatha yokuba sengozini ngolimi olulula futhi bathole izimpendulo ezisheshayo ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). I-ExposureAI "ingafingqa indlela ephelele yokuhlasela endabeni" ngobungozi obunikeziwe obubalulekile, ichaza ukuthi umhlaseli angayibopha kanjani ngobunye ubuthakathaka ukuze afake isistimu engozini. Ize iphakamise nezenzo zokulungisa futhi iphendule imibuzo yokulandelela mayelana nobungozi. Lokhu kusho ukuthi lapho kumenyezelwa i-CVE entsha ebucayi (Ukuba Sengozini Okujwayelekile Nokuchayeka), umhlaziyi angabuza i-AI, “Ingabe kukhona amaseva ethu athintwe yile CVE futhi yisiphi isimo esibi kakhulu uma singaqedi?” futhi uthole ukuhlolwa okucacile okuthathwe kudatha yokuskena yenhlangano. Ngokwenza ubungozi bomongo (isb. lena ivezwa ku-inthanethi nakuseva yenani eliphezulu, ngakho-ke ibaluleke kakhulu), i-AI ekhiqizayo isiza amaqembu ukuthi axhumane ngobuhlakani nezinsiza ezilinganiselwe.

Ngokungeziwe ekutholeni nasekulawuleni ubungozi obaziwayo, i-AI ekhiqizayo inikela ekuhlolweni kokungena nasekulingiseni ukuhlasela - ngokuyisisekelo ukuthola obungaziwa noma izilawuli zokuphepha zokuhlola. Amanethiwekhi ezitha ezikhiqizayo (ama-GAN), uhlobo lokukhiqiza lwe-AI, asetshenziselwe ukudala idatha yokwenziwa elingisa ithrafikhi yenethiwekhi yangempela noma ukuziphatha komsebenzisi, okungabandakanya amaphethini okuhlasela afihliwe. Ucwaningo lwango-2023 luphakamise ukusebenzisa ama-GAN ukuze kukhiqizwe ithrafikhi yokuhlasela yosuku oluyiziro engokoqobo ukuze kuqeqeshwe amasistimu okuthola ukungena ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Ngokuphakela i-IDS ngezimo zokuhlasela eziklanywe yi-AI (ezingafaki engozini ukusebenzisa uhlelo olungayilungele ikhompuyutha langempela kumanethiwekhi okukhiqiza), izinhlangano zingaqeqesha ukuzivikela kwazo ukuze zibone izinsongo ezintsha ngaphandle kokulinda ukushaywa yibona empeleni. Ngokufanayo, i-AI ingakwazi ukulingisa umhlaseli ohlola uhlelo - ngokwesibonelo, ezama ngokuzenzakalelayo amasu ahlukahlukene okuxhaphaza endaweni ephephile ukuze abone ukuthi akhona yini aphumelelayo. I-US Defence Advanced Research Projects Agency (DARPA) ibona isithembiso lapha: i-2023 AI Cyber Challenge yayo isebenzisa ngokucacile i-AI ekhiqizayo (njengamamodeli amakhulu olimi) ukuze “ithole futhi ilungise ngokuzenzekelayo ubungozi kusofthiwe yomthombo ovulekile” njengengxenye yomncintiswano ( I-DARPA Ihlose Ukuthuthukisa I-AI, Izicelo Zokuzimela, I-Autonomy Applications of Defence > Umnyango Wezokuvikela wase-US Ungathembela ). Lesi sinyathelo sigcizelela ukuthi i-AI ayisizi nje ukuhlanganisa izimbobo ezaziwayo; yembula okusha futhi iphakamisa ukulungiswa, umsebenzi ngokuvamile okhawulelwe kubacwaningi bezokuphepha abanekhono (nababizayo).

I-Generative AI ingakha ngisho nezimbiza zezinyosi ezihlakaniphile namawele edijithali okuzivikela. Ama-Startups athuthukisa amasistimu wokukhohlisa aqhutshwa yi-AI alingisa ngokugculisayo amaseva wangempela noma amadivayisi. Njengoba i-CEO eyodwa yachaza, i-AI ekhiqizayo ingakwazi "ukuhlanganisa izinhlelo zedijithali ukuze zilingise ezangempela futhi zihehe abaduni" ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Lezi zimbiza zoju ezikhiqizwe yi-AI ziziphatha njengendawo ezungezile (ake sithi, idivayisi ye-IoT mbumbulu ethumela i-telemetry evamile) kodwa zikhona ukuze zihehe abahlaseli. Lapho umhlaseli eqondise ekukhohlisweni, i-AI ibakhohlisile ukuthi baveze izindlela zabo, abavikeli abangase bazifunde futhi bazisebenzise ukuqinisa izinhlelo zangempela. Lo mqondo, onikwe amandla ukumodela okukhiqizayo, uhlinzeka ngendlela ebheke phambili yokuvula amatafula kubahlaseli , kusetshenziswa ukukhohlisa okuthuthukiswe yi-AI.

Kuzo zonke izimboni, ukuphatha ukuba sengozini okusheshayo nokuhlakaniphile kusho ukwephulwa kwemithetho embalwa. Ekunakekelweni kwezempilo kwe-IT, ngokwesibonelo, i-AI ingase ibone ngokushesha umtapo wolwazi osengcupheni esisetshenziswa sezokwelapha futhi yazise ukulungiswa kwe-firmware ngaphambi kokuba noma yimuphi umhlaseli ayisebenzise. Ebhange, i-AI ingalingisa ukuhlasela kwangaphakathi kuhlelo olusha lokuqinisekisa ukuthi idatha yekhasimende ihlala iphephile ngaphansi kwazo zonke izimo. Ngakho-ke i-Generative AI isebenza njengesibonakhulu kanye nesihloli sokucindezeleka sokuma kokuphepha kwezinhlangano: ikhanyisa amaphutha afihliwe futhi icindezela izinhlelo ngezindlela ezicatshangelwayo zokuqinisekisa ukuqina.

Vikela Ukukhiqizwa Kwekhodi Nokuthuthukiswa Kwesoftware

Amakhono e-Generative AI awagcini nje ekutholeni ukuhlaselwa - adlulela nasekudaleni amasistimu avikeleke kakhulu kusukela ekuqaleni . Ekuthuthukisweni kwesofthiwe, amajeneretha ekhodi ye-AI (njenge-GitHub Copilot, i-OpenAI Codex, njll.) angasiza onjiniyela babhale ikhodi ngokushesha ngokuphakamisa amazwibela ekhodi noma yonke imisebenzi. I-engeli ye-cybersecurity iqinisekisa ukuthi lezi zingcezu zekhodi eziphakanyiswe yi-AI zivikelekile futhi zisebenzisa i-AI ukuthuthukisa izinqubo zokubhala amakhodi.

Ngakolunye uhlangothi, i-AI ekhiqizayo ingasebenza njengomsizi wokubhala amakhodi oshumeka izinqubo ezihamba phambili zokuphepha . Onjiniyela bangakwazi ukwazisa ithuluzi le-AI, “Khiqiza umsebenzi wokusetha kabusha iphasiwedi ku-Python,” futhi ngokufanelekile bathole ikhodi engasebenzi kuphela kodwa futhi elandela imihlahlandlela evikelekile (isb. ukuqinisekiswa kokufakwayo okufanele, ukugawula, ukuphatha amaphutha ngaphandle kolwazi oluvuzayo, njll.). Umsizi onjalo, oqeqeshwe ngezibonelo eziningi zekhodi evikelekile, angasiza ekunciphiseni amaphutha abantu aholela ekubeni sengozini. Isibonelo, uma unjiniyela ekhohlwa ukuhlanza okokufaka komsebenzisi (evula umnyango womjovo we-SQL noma izinkinga ezifanayo), i-AI ingase ifake lokho ngokuzenzakalelayo noma ibaxwayise. Amanye amathuluzi okubhala amakhodi we-AI manje alungiswa kahle ngedatha egxile ekuvikelekeni ukuze afeze le njongo - empeleni, ukuhlela kwe-AI okuhambisana nonembeza wokuphepha .

Kodwa-ke, kukhona uhlangothi: i-AI ekhiqizayo ingakwazi ukwethula kalula ubungozi uma ingalawulwa ngendlela efanele. Njengoba uchwepheshe wezokuphepha we-Sophos u-Ben Verschaeren aphawula, ukusebenzisa i-AI ekhiqizayo ekubhaleni amakhodi “kulungile ngekhodi emfushane, eqinisekisekayo, kodwa kuyingozi lapho ikhodi engahloliwe ihlanganiswa” ezinhlelweni zokukhiqiza. Ingozi ukuthi i-AI ingase ikhiqize ikhodi enengqondo engavikelekile ngezindlela umuntu ongeyena uchwepheshe angase angaboni. Ngaphezu kwalokho, abalingisi abanonya bangathonya ngamabomu amamodeli e-AI omphakathi ngokuwatshala ngamaphethini ekhodi asengozini (uhlobo lobuthi bedatha) ukuze i-AI iphakamise ikhodi engavikelekile. Onjiniyela abaningi ababona ochwepheshe bezokuphepha , ngakho-ke uma i-AI iphakamisa isixazululo esikahle, bangase bayisebenzise ngokunganaki, bengaboni ukuthi inesici ( 6 Sebenzisa Amakesi E-Generative AI ku-Cybersecurity [+ Izibonelo] ). Lokhu kukhathazeka kungokoqobo - empeleni, kunohlu lwe-OWASP Top 10 manje lwama-LLM (amamodeli ezilimi ezinkulu) oluveza ubungozi obuvamile njengalokhu ekusebenziseni i-AI ekubhaleni amakhodi.

Ukubhekana nalezi zinkinga, ochwepheshe baphakamisa "ukulwa ne-AI ekhiqizayo nge-generative AI" endaweni yokubhala ikhodi. Empeleni, lokho kusho ukusebenzisa i-AI ukuze ubuyekeze futhi uhlole ikhodi eyabhalwa enye i-AI (noma abantu). I-AI ingakwazi ukuskena ngekhodi entsha yenza ngokushesha kakhulu kunombuyekezi wekhodi yomuntu futhi ihlabe umkhosi ubungozi obungaba khona noma izinkinga ezinengqondo. Sesivele siwabona amathuluzi avelayo ahlanganisa umjikelezo wokuphila wokuthuthukiswa kwesofthiwe: ikhodi ibhaliwe (mhlawumbe ngosizo lwe-AI), bese imodeli ekhiqizayo eqeqeshwe ngezimiso zekhodi evikelekile iyayibuyekeza futhi ikhiphe umbiko wanoma yikuphi ukukhathazeka (ithi, ukusetshenziswa kwemisebenzi eyehlisiwe, ukuhlola ukuqinisekiswa kobuqiniso okulahlekile, njll.). Ucwaningo lwe-NVIDIA, okukhulunywe ngalo ekuqaleni, oluthole ukutholakala kobungozi obungu-4× ngokushesha kukhodi luyisibonelo sokusebenzisa i-AI ukuze kuhlaziywe ikhodi evikelekile ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ).

Ngaphezu kwalokho, i-AI ekhiqizayo ingasiza ekudaleni ukucushwa okuphephile nemibhalo . Isibonelo, uma inkampani idinga ukuphakela ingqalasizinda yefu evikelekile, unjiniyela angacela i-AI ukuthi yenze imibhalo yokucushwa (Ingqalasizinda njengeKhodi) enezilawuli zokuphepha (njengokuhlukaniswa kwenethiwekhi okufanele, izindima ze-IAM ezinelungelo elincane) ezibhakwe. I-AI, njengoba iqeqeshwe ezinkulungwaneni zokucushwa okunjalo, ingakhiqiza isisekelo esizokwenziwa unjiniyela. Lokhu kusheshisa ukusethwa okuvikelekile kwamasistimu futhi kunciphisa amaphutha okulungisa - umthombo ovamile wezigameko zokuphepha kwamafu.

Ezinye izinhlangano ziphinde zisebenzisa i-AI ekhiqizayo ukuze kugcinwe isisekelo solwazi samaphethini wokubhala amakhodi avikelekile. Uma unjiniyela engenasiqiniseko sokuthi usifaka kanjani isici esithile ngokuphephile, angabuza nge-AI yangaphakathi efunde kumaphrojekthi enkampani adlule nemihlahlandlela yezokuphepha. I-AI ingase ibuyisele indlela enconyiwe noma amazwibela ekhodi ahambisana nazo zombili izidingo zokusebenza kanye namazinga okuphepha enkampani. Le ndlela isetshenziswe ngamathuluzi afana ne- Secureframe's Questionnaire Automation , edonsa izimpendulo kuzinqubomgomo zenkampani kanye nezixazululo zesikhathi esidlule ukuze kuqinisekiswe izimpendulo ezingaguquki nezinembile (empeleni zikhiqiza imibhalo evikelekile) ( I-Generative AI Ingasetshenziswa Kanjani Ku-Cybersecurity? Izibonelo Zezwe Langempela Ezingu-10 ). Umqondo uhumusheka ekubhaleni ikhodi: i-AI “ekhumbula” ukuthi usebenzise kanjani okuthile ngokuphephile ngaphambili futhi ikuqondise ukuthi ukwenze ngaleyo ndlela futhi.

Kafushane, i-AI ekhiqizayo inomthelela ekuthuthukisweni kwesoftware ngokwenza usizo oluvikelekile lokufaka amakhodi lufinyeleleke kakhudlwana . Izimboni ezithuthukisa ama-software amaningi angokwezifiso - ezobuchwepheshe, ezezimali, ezokuvikela, njll. - zizozuza ngokuba namakhophi e-AI angasheshi nje ukubhala amakhodi kodwa asebenze njengombuyekezi wezokuphepha ohlale eqaphile. Uma elawulwa kahle, lawa mathuluzi e-AI anganciphisa ukwethulwa kobungozi obusha futhi asize amaqembu athuthukisisayo ukuthi abambelele ezenzweni ezingcono kakhulu, ngisho noma iqembu lingenaye uchwepheshe wezokuphepha ohilelekile kuzo zonke izinyathelo. Umphumela uba isofthiwe enamandla kakhulu ngokumelene nokuhlaselwa kusukela ngosuku lokuqala.

Ukwesekwa Kwempendulo Yesigameko

Uma kwenzeka isigameko sokuvikeleka ku-inthanethi - kungaba ukuqubuka kwe-malware, ukwephulwa kwedatha, noma ukuphela kwesistimu ekuhlaselweni - isikhathi sibalulekile. I-Generative AI isetshenziswa ngokwandayo ukuze kusekelwe amaqembu okusabela ngesigameko (IR) ekuhlanganiseni nasekulungiseni izehlakalo ngokushesha nangolwazi olwengeziwe oluseduze. Umqondo wukuthi i-AI ingakwazi ukuthwala omunye umthwalo wophenyo kanye nemibhalo ngesikhathi sesigameko, futhi iphakamise noma izenzele ngokuzenzakalelayo ezinye izenzo zokuphendula.

Indima eyodwa ebalulekile ye-AI ku-IR ukuhlaziya kanye nokufingqwa kwesigameko sesikhathi sangempela . Phakathi nesigameko, abaphenduli bangase badinge izimpendulo zemibuzo efana nokuthi "Umhlaseli ungene kanjani?" , "Iziphi izinhlelo ezithintekayo?" , kanye nokuthi “Iyiphi idatha engase ibe sengozini?” . I-Generative AI ingahlaziya amalogi, izexwayiso, kanye nedatha ye-forensic evela kumasistimu athintekile futhi inikeze imininingwane ngokushesha. Isibonelo, i-Microsoft Security Copilot ivumela umphenduli wesigameko ukuthi afunze ngezingxenye ezihlukahlukene zobufakazi (amafayela, ama-URL, amalogi omcimbi) futhi acele umugqa wesikhathi noma isifinyezo ( I-Microsoft Security Copilot iwumsizi omusha we-GPT-4 AI we-cybersecurity | The Verge ). I-AI ingase iphendule ngokuthi: “Ukwephulwa komthetho kungenzeka kuqale nge-imeyili yobugebengu bokweba imininingwane ebucayi eya kumsebenzisi u-JohnDoe ngo-10:53 GMT equkethe i-malware X. Uma isikhishiwe, uhlelo olungayilungele ikhompuyutha ludale i-backdoor eyasetshenziswa ezinsukwini ezimbili kamuva ukuhambisa eceleni kuseva yezezimali, lapho yaqoqa khona idatha.” Ukuba nalesi sithombe esihambisanayo ngemizuzu esikhundleni samahora kwenza ithimba likwazi ukwenza izinqumo ezinolwazi (njengokuthi yiziphi izinhlelo okufanele zihlukanise) ngokushesha okukhulu.

I-Generative AI ingase futhi iphakamise izenzo zokuqukatha nokulungisa . Isibonelo, uma isiphetho singenwe yi-ransomware, ithuluzi le-AI lingakwazi ukukhiqiza iskripthi noma isethi yemiyalo yokuhlukanisa lowo mshini, ikhubaze ama-akhawunti athile, futhi ivimbe ama-IP ayingozi aziwayo ku-firewall - empeleni ukusetshenziswa kwe-playbook. I-Palo Alto Networks iphawula ukuthi i-AI ekhiqizayo iyakwazi "ukukhiqiza izenzo ezifanele noma izikripthi ezisekelwe esimweni sesigameko" , ngokuzenzakalelayo izinyathelo zokuqala zokuphendula ( Iyini i-Generative AI ku-Cybersecurity? - I-Palo Alto Networks ). Esimeni lapho ithimba lokuvikela likhungathekile (isho ukuhlasela okusabalele kuwo wonke amadivayisi), i-AI ingase yenze ngokuqondile ezinye zalezi zenzo ngaphansi kwezimo ezigunyazwe ngaphambilini, isebenze njengesiphenduli esincane esisebenza ngokungakhathali. Isibonelo, umenzeli we-AI angasetha kabusha ngokuzenzakalelayo izifakazelo abona sengathi sengozini noma avalele abasingathi ababonisa umsebenzi omubi ofana nephrofayela yesigameko.

Ngesikhathi sokuphendula ngesigameko, ukuxhumana kubalulekile - ngaphakathi kweqembu kanye nababambe iqhaza. I-Generative AI ingasiza ngokubhala imibiko yesibuyekezo sesigameko noma amafuphi ngokuhamba kwesikhathi . Esikhundleni sokuthi unjiniyela amise ukuxazulula izinkinga zakhe ukuze abhale isibuyekezo se-imeyili, angabuza i-AI, "Fingqa okwenzekile kulesi sigameko kuze kube manje ukuze wazise abaphathi." I-AI, njengoba ingenise idatha yesigameko, ingaveza isifinyezo esifushane: "Kusukela ngo-3 PM, abahlaseli bafinyelele ama-akhawunti abasebenzisi angu-2 namaseva angu-5. Idatha ethintekile ihlanganisa amarekhodi eklayenti kusizindalwazi X. Izinyathelo zokuqukethwe: Ukufinyelela kwe-VPN kwama-akhawunti onakalisiwe kuhoxisiwe futhi amaseva ahlukanisiwe. Izinyathelo ezilandelayo: ukuskena noma yiziphi izindlela zokuphikelela." Ophendulayo angaqinisekisa ngokushesha noma alungise lokhu futhi akuthumele, aqinisekise ukuthi ababambiqhaza bagcinwa benolwazi olunembile, olufika kumzuzu.

Ngemuva kokuthi uthuli seluthulile, kuvame ukuba nombiko onemininingwane yesigameko okumele ulungiselelwe kanye nezifundo ezifundiwe ukuzihlanganisa. Lena enye indawo lapho ukusekelwa kwe-AI kukhanya khona. Ingabuyekeza yonke idatha yesigameko futhi ikhiqize umbiko wangemva kwesigameko ohlanganisa imbangela, ukulandelana kwezikhathi, umthelela, nezincomo. I-IBM, ngokwesibonelo, ihlanganisa i-AI ekhiqizayo ukuze idale "izifinyezo ezilula zamacala okuphepha nezigameko ezingabiwa nabathintekayo" ngokucindezela inkinobho ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? 10 Izibonelo Zomhlaba Wangempela ). Ngokuhlela ukubika ngemva kwesenzo, izinhlangano zingasebenzisa ukuthuthukiswa ngokushesha futhi zibe namadokhumenti angcono ngezinjongo zokuthobelana.

Ukusetshenziswa okukodwa okubheka phambili okuthuthukile ukulingisa izigameko eziqhutshwa yi-AI . Ngokufana nendlela umuntu angasebenzisa ngayo umshini wokuzicisha umlilo, ezinye izinkampani zisebenzisa i-AI yokukhiqiza ukuze zibhekane nezimo zezehlakalo “zokuthiwani-uma”. I-AI ingase ilingise ukuthi i-ransomware ingase isakazeke kanjani uma kubhekwa isakhiwo senethiwekhi, noma ukuthi umuntu wangaphakathi angayikhipha kanjani idatha, bese ethola ukuphumelela kwezinhlelo zamanje zokuphendula. Lokhu kusiza amaqembu ukuthi alungiselele futhi acwenge izincwadi zokudlala ngaphambi kokuba kwenzeke isigameko sangempela. Kufana nokuba nomeluleki wempendulo yesigameko ohlale ethuthuka ohlola ukuthi ukulungele yini.

Ezimbonini ezisezingeni eliphezulu njengezezimali noma ezokunakekelwa kwempilo, lapho isikhathi sokuphumula noma ukulahleka kwedatha kusuka ezehlakalweni kubiza kakhulu, lawa makhono e-IR aqhutshwa yi-AI akhanga kakhulu. Isibhedlela esinesigameko se-inthanethi asikwazi ukukhokhela ukuphela kwesistimu isikhathi eside - i-AI esiza ngokushesha ekunqandeni ingase isindise ukuphila. Ngokufanayo, isikhungo sezezimali singasebenzisa i-AI ukuze sisingathe isivivinyo sokuqala sokungena kokukhwabanisa okusolisayo ngo-3 AM, ukuze ngesikhathi abantu abasocingweni sebeku-inthanethi, kunesisekelo esiningi (ukukhipha ama-akhawunti athintekile, ukuvinjwa kokuthengiswa, njll.) sekuvele kwenziwa. Ngokwandisa amaqembu okuphendula izigameko nge-AI ekhiqizayo , izinhlangano zinganciphisa kakhulu izikhathi zokuphendula futhi zithuthukise ukunemba kokuphatha kwazo, ekugcineni zinciphise umonakalo ovela ezigamekweni ze-cyber.

Izibalo Zokuziphatha kanye Nokutholwa Okungaqondakali

Ukuhlaselwa ku-inthanethi okuningi kungabanjwa ngokuqaphela lapho okuthile kuphambuka ekuziphatheni “okuvamile” – kungakhathaliseki ukuthi i-akhawunti yomsebenzisi elanda inani elingajwayelekile ledatha noma idivayisi yenethiwekhi ngokuzumayo ixhumana nomsingathi ongamjwayele. I-Generative AI inikezela ngamasu athuthukile okuhlaziya ukuziphatha kanye nokutholwa okudidayo , ukufunda amaphethini avamile wabasebenzisi namasistimu bese imaka uma okuthile kungabonakali.

Ukutholwa okudidayo okujwayelekile kuvame ukusebenzisa izilinganiso zezibalo noma ukufunda komshini okulula kumamethrikhi athile (ama-CPU okusetshenziswa kwe-spikes, ukungena ngemvume ngamahora angajwayelekile, njll.). I-Generative AI ingaqhubekisela phambili lokhu ngokwakha amaphrofayli amaningi okuziphatha. Isibonelo, imodeli ye-AI ingangenisa ukungena ngemvume, amaphethini okufinyelela amafayela, nemikhuba ye-imeyili yesisebenzi ngokuhamba kwesikhathi futhi yakhe ukuqonda okunezinhlangothi eziningi “kokuvamile” kwalowo msebenzisi. Uma kamuva leyo akhawunti yenza okuthile okungaphezu kokujwayelekile kwayo (njengokungena ngemvume usuka ezweni elisha nokufinyelela inqwaba yamafayela e-HR phakathi kwamabili), i-AI izobona ukuchezuka hhayi kumethrikhi eyodwa kuphela kodwa yonke iphethini yokuziphatha engahambisani nephrofayela yomsebenzisi. Ngamagama obuchwepheshe, amamodeli akhiqizayo (njengama-autoencoder noma amamodeli alandelanayo) angenza imodeli yokuthi "okuvamile" kubukeka kanjani bese ekhiqiza ububanzi bokuziphatha okulindelekile. Uma okungokoqobo kuwela ngaphandle kwalolo bubanzi, kumakwa njengokudidayo ( What Is Generative AI in Cybersecurity? - Palo Alto Networks ).

Ukuqaliswa okukodwa okungokoqobo kusekugadweni kwethrafikhi yenethiwekhi . Ngokwenhlolovo ye-2024, i-54% yezinhlangano zase-US zicaphune ukuqapha ithrafikhi yenethiwekhi njengento ephezulu yokusetshenziswa kwe-AI ku-cybersecurity ( I-North America: amacala aphezulu okusebenzisa i-AI ku-cybersecurity emhlabeni wonke i-2024 ). I-Generative AI ingafunda amaphethini okuxhumana avamile enethiwekhi yebhizinisi – okuyiziphakeli ngokuvamile ezikhuluma zodwa, ukuthi yimaphi amavolumu edatha ahambayo phakathi namahora ebhizinisi uma kuqhathaniswa ngobusuku obubodwa, njll. Uma umhlaseli eqala ukuhlunga idatha esuka kuseva, ngisho kancane ukuze agweme ukutholwa, isistimu esekelwe ku-AI ingase iqaphele ukuthi “Iseva A ayithumeli u-500MB wedatha ku-IPrt yangaphandle” futhi iphakamise i-IPrt yangaphandle. Ngenxa yokuthi i-AI ayisebenzisi nje imithetho emile kodwa iyimodeli ethuthukayo yokuziphatha kwenethiwekhi, ingathola okungaqondakali okunemithetho emile (njengokuthi “ukwazisa uma idatha > X MB”) ingase igeje noma ihlabe umkhosi ngephutha. Lesi simo esiguquguqukayo yisona esenza ukutholakala okudidayo okushayelwa yi-AI kube namandla ezindaweni ezifana namanethiwekhi okwenziwa kwamabhange, ingqalasizinda yamafu, noma izindiza zedivayisi ye-IoT, lapho ukuchaza imithetho emisiwe yokujwayelekile nokungajwayelekile kuyinkimbinkimbi kakhulu.

I-Generative AI iphinde isize ngezibalo zokuziphatha komsebenzisi (UBA) , okuwukhiye wokubona izinsongo zangaphakathi noma ama-akhawunti onakalisiwe. Ngokukhiqiza isisekelo somsebenzisi ngamunye noma ibhizinisi, i-AI ingathola izinto ezifana nokusetshenziswa kabi kwemininingwane. Isibonelo, uma u-Bob ovela ku-accounting ngokuzumayo eqala ukubuza kusizindalwazi samakhasimende (into angakaze ayenze ngaphambilini), imodeli ye-AI yokuziphatha kuka-Bob izomaka lokhu njengokungajwayelekile. Kungase kungabi uhlelo olungayilungele ikhompuyutha - kungaba indaba yokuthi imininingwane ka-Bob intshontshiwe futhi isetshenziswe umhlaseli, noma u-Bob ephenya lapho kungafanele khona. Kunoma ikuphi, ithimba lezokuphepha liyatshelwa ukuze liphenye. Amasistimu anjalo e-UBA aqhutshwa yi-AI akhona emikhiqizweni ehlukahlukene yezokuvikela, futhi amasu okumodela akhiqizayo aphushela ukunemba kwawo phezulu futhi ehlisa ama-alamu angamanga ngokucabangela umongo (mhlawumbe u-Bob ukuphrojekthi ekhethekile, njll., i-AI ngezinye izikhathi engakwazi ukuyichaza ivela kwenye idatha).

Emkhakheni wobunikazi kanye nokuphathwa kokufinyelela, ukutholwa kwe-deepfake kuyisidingo esikhulayo - i-AI ekhiqizayo ingakha amazwi okwenziwa namavidiyo akhohlise ukuphepha kwe-biometric. Kuyathakazelisa ukuthi i-AI ekhiqizayo ingasiza futhi ukuthola lawa ma-deepfakes ngokuhlaziya izinto zobuciko ezicashile kumsindo noma kuvidiyo okunzima ukuthi abantu bakubone. Sibone isibonelo nge-Accenture, esebenzisa i-AI ekhiqizayo ukuze ilingise isimo sobuso nezimo ezingenakubalwa ukuqeqesha amasistimu abo ebhayomethrikhi ukuze ahlukanise abasebenzisi bangempela kuma-deepfakes akhiqizwe yi-AI. Eminyakeni emihlanu, le ndlela yasiza i-Accenture ukuqeda amaphasiwedi we-90% yezinhlelo zayo (ukuthuthela ku-biometrics nezinye izici) futhi yehlise ukuhlaselwa ngo-60% ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Empeleni, basebenzise i-AI ekhiqizayo ukuqinisa ukuqinisekiswa kwebhayomethrikhi, beyenza ikwazi ukumelana nokuhlaselwa okuzalayo (umfanekiso omkhulu we-AI yokulwa ne-AI). Lolu hlobo lwemodeli yokuziphatha - kulokhu ukubona umehluko phakathi kobuso bomuntu obuphilayo ngokumelene nobunye obuhlanganiswe ne-AI - lubalulekile njengoba sithembele kakhulu ku-AI ekuqinisekiseni ubuqiniso.

Ukutholwa okudidayo okunikwa amandla yi-generative AI kusebenza kuzo zonke izimboni: kwezokunakekelwa kwezempilo, ukuqapha ukuziphatha kwedivayisi yezokwelapha ukuze uthole izimpawu zokugetshengwa; kwezezimali, ukubukela amasistimu okuhweba amaphethini angajwayelekile angabonisa ukukhwabanisa noma ukukhohlisa kwe-algorithmic; emandleni/ezinsizakalweni, ukubheka amasiginali wokulawula wezimpawu zokugxambukela. Inhlanganisela yobubanzi (ukubheka zonke izici zokuziphatha) nokujula (ukuqonda amaphethini ayinkimbinkimbi) ehlinzekwa yi-AI ekhiqizayo kuyenza ibe ithuluzi elinamandla lokubona izinkomba zenaliti-in-a-haystack zesigameko se-cyber. Njengoba izinsongo ziba ngobuqili, zicasha phakathi kwemisebenzi evamile, leli khono lokubonisa ngokuqondile "okuvamile" futhi umemeze uma okuthile kuchezuka kubaluleka. Ngakho-ke i-Generative AI isebenza njengomthumeli ongakhathali, ehlala efunda futhi ebuyekeza incazelo yayo yokujwayelekile ukuze ihambisane nezinguquko zemvelo, futhi ixwayise amaqembu ezokuphepha ngezinto ezididayo okufanele zihlolwe ngokucophelela.

Amathuba kanye nezinzuzo ze-Generative AI ku-Cybersecurity

Ukusetshenziswa kwe-AI ekhiqizayo ku-cybersecurity kuletha inqwaba yamathuba nezinzuzo zezinhlangano ezizimisele ukwamukela lawa mathuluzi. Ngezansi, sifingqa izinzuzo ezibalulekile ezenza i-AI ekhiqizayo ibe isengezo esiphoqayo ezinhlelweni zokuphepha ku-inthanethi:

Ukutholwa Okusheshayo Nokusabela Kosongo: Amasistimu e-AI akhiqizayo angahlaziya inani elikhulu ledatha ngesikhathi sangempela futhi abone izinsongo ngokushesha kunokuhlaziya komuntu mathupha. Le nzuzo yesivinini isho ukutholwa ngaphambi kwesikhathi kokuhlaselwa kanye nokunqanda isigameko esisheshayo. Empeleni, ukuqapha kwezokuphepha okuqhutshwa yi-AI kungabamba izinsongo ezingathatha abantu isikhathi eside ukuhambisana. Ngokuphendula izehlakalo ngokushesha (noma ngisho nokwenza ngokuzenzakalelayo izimpendulo zokuqala), izinhlangano zinganciphisa kakhulu isikhathi sokuhlala sabahlaseli kumanethiwekhi azo, zinciphise umonakalo.
Ukunemba Okuthuthukisiwe Nokwembozwa Kosongo: Ngenxa yokuthi afunda ngokuqhubekayo kudatha entsha, amamodeli akhiqizayo angakwazi ukuzivumelanisa nezinsongo eziguqukayo futhi abambe izimpawu ezicashile zomsebenzi onobungozi. Lokhu kuholela ekutholeni ukunemba okuthuthukisiwe (izinto ezimbalwa ezingezinhle ezingamanga nezingezinhle ezingamanga) uma kuqhathaniswa nemithetho emile. Isibonelo, i-AI efunde izimpawu ze-imeyili yobugebengu bokweba imininingwane ebucayi noma ukuziphatha kohlelo olungayilungele ikhompuyutha ingakhomba okuhlukile okungakaze kubonwe ngaphambilini. Umphumela uwukumbozwa okubanzi kwezinhlobo ezisabisayo - okuhlanganisa nokuhlasela okunoveli - okuqinisa ukuma kwezokuphepha kukonke. Amathimba okuvikela aphinde athole imininingwane enemininingwane ekuhlaziyweni kwe-AI (isb. izincazelo zohlelo olungayilungele ikhompuyutha), okunika amandla ukuzivikela okunembe kakhudlwana nokuqondiwe ( What Is Generative AI in Cybersecurity? - Palo Alto Networks ).
Ukuzenzakalela Kwemisebenzi Ephindaphindwayo: I-Generative AI iphumelela kakhulu enqubweni yokuzenzakalela, imisebenzi yezokuphepha edinga abasebenzi abaningi - kusukela ekuhlanganiseni izingodo nokuhlanganisa imibiko kuya ekubhaleni imibhalo yempendulo yesigameko. Lokhu kuzenzakalela kunciphisa umthwalo kubahlaziyi abangabantu , kubakhulula ukuba bagxile kumasu wezinga eliphezulu kanye nokwenza izinqumo eziyinkimbinkimbi ( Iyini i-Generative AI ku-Cybersecurity? - I-Palo Alto Networks ). Imisebenzi evamile kodwa ebalulekile njengokuskena kokuba sengozini, ukuhlolwa kokuhlelwa, ukuhlaziya umsebenzi wabasebenzisi, nokubikwa kokuthobelana kungasingathwa (noma okungenani okubhalwe kuqala) yi-AI. Ngokuphatha le misebenzi ngesivinini somshini, i-AI ayigcini nje ngokuthuthukisa ukusebenza kahle kodwa futhi inciphisa amaphutha omuntu (isici esibalulekile ekuphuleni umthetho).
Ukuvikela Okusebenzayo Nokulingisa: I-Generative AI ivumela izinhlangano ukuthi zisuke ekusebenzeni ziye ekuvikelekeni okusebenzayo. Ngamasu afana nokulingisa ukuhlasela, ukukhiqizwa kwedatha yokwenziwa, nokuqeqeshwa okusekelwe kusimo, abavikeli bangalindela futhi balungiselele izinsongo ngaphambi kokuthi zenzeke emhlabeni wangempela. Amathimba okuvikela angakwazi ukulingisa ukuhlasela kwe-inthanethi (imikhankaso yobugebengu bokweba imininingwane ebucayi, ukuqubuka kwe-malware, i-DDoS, njll.) endaweni ephephile ukuze ahlole izimpendulo zawo futhi aqinise noma yibuphi ubuthakathaka. Lokhu kuqeqeshwa okuqhubekayo, ngokuvamile okungenzeki kahle ngomzamo womuntu nje, kugcina izivikelo zibukhali futhi zihambisana nesikhathi. Kufana “nomshini wokuprakthiza umlilo” we-cyber - I-AI ingaphonsa izinsongo eziningi ezicatshangelwayo ekuzivikeleni kwakho ukuze uzilolonge futhi uthuthuke.
I-Augmenting Human Expertise (AI njenge-Force Multiplier): I-Generative AI isebenza njengomhlaziyi omncane ongakhathali, umeluleki, nomsizi ogingqelwe kokukodwa. Inganikeza amalungu eqembu angenalwazi olutheni ngesiqondiso nezincomo ngokuvamile okulindeleke kochwepheshe abangomakad’ ebona, uchwepheshe obusa intando yeningi eqenjini lonke ( 6 Use Cases for Generative AI in Cybersecurity [+ Examples] ). Lokhu kubaluleke kakhulu uma kubhekwa ukushoda kwethalente ku-cybersecurity - i-AI isiza amaqembu amancane ukuthi enze okuningi ngokuncane. Ngakolunye uhlangothi, abahlaziyi abanolwazi bayahlomula emsebenzini we-AI wokusingatha i-grunt kanye nemininingwane engabonakali, abangase bayiqinisekise futhi benze ngayo. Umphumela usuwonke uyithimba lezokuphepha elikhiqiza kakhulu futhi elikwaziyo, nge-AI ekhulisa umthelela welungu ngalinye elingumuntu ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity ).
Ukusekelwa Kwesinqumo Okuthuthukisiwe Nokubikwa: Ngokuhumusha idatha yobuchwepheshe ibe imininingwane yolimi lwemvelo, i-AI ekhiqizayo ithuthukisa ukuxhumana nokwenza izinqumo. Abaholi bezokuphepha bathola ukubonakala okucacile ezindabeni ngokusebenzisa izifinyezo ezenziwe nge-AI futhi bangenza izinqumo zamasu ezinolwazi ngaphandle kokudinga ukuhlaziya idatha eluhlaza. Ngokufanayo, ukuxhumana okuhlukahlukene (kubaphathi, izikhulu zokuthobela, njll.) kuyathuthukiswa lapho i-AI ilungiselela imibiko eqondakala kalula yokuma kwezokuphepha kanye nezigameko ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Zezwe Langempela Ezingu-10 ). Lokhu akwakhi nje ukuzethemba nokuqondanisa ezindabeni zokuphepha ezingeni lobuholi kodwa futhi kusiza ukuthethelela ukutshalwa kwezimali nezinguquko ngokuveza ngokucacile ubungozi kanye nezikhala ezitholwe yi-AI.

Ngokuhlangene, lezi zinzuzo zisho ukuthi izinhlangano ezisebenzisa i-AI ekhiqizayo ku-cybersecurity zingafinyelela ukuma okuqinile kokuphepha ngezindleko zokusebenza okungenzeka zibe phansi. Bangasabela ezinsongweni ebezingaphezu kwamandla phambilini, bavale izikhala ezingazange ziqashwe, futhi bathuthuke ngokuqhubekayo ngamalophu empendulo aqhutshwa yi-AI. Ekugcineni, i-AI ekhiqizayo inikeza ithuba lokudlula izitha ngokufanisa isivinini, isikali, nobunkimbinkimbi bokuhlasela kwesimanje ngokuzivikela okuyinkimbinkimbi ngokulinganayo. Njengoba enye inhlolovo yathola, ngaphezu kwesigamu sabaholi bamabhizinisi nabe-inthanethi balindele ukutholwa kosongo ngokushesha kanye nokunemba okwandisiwe ngokusebenzisa i-AI ekhiqizayo ( [PDF] I-Global Cybersecurity Outlook 2025 | I-World Economic Forum ) ( I-Generative AI ku-Cybersecurity: Ukubuyekeza Okubanzi kwe-LLM ... ) - ubufakazi bethemba elihambisana nalobu buchwepheshe.

Izingozi Nezinselele Zokusebenzisa I-Generative AI ku-Cybersecurity

Yize amathuba ebalulekile, kubalulekile ukusondela ku-AI ekhiqizayo ekuvikelekeni kwe-inthanethi ngamehlo avulekele ubungozi nezinselelo ezihilelekile. Ukuthembela ngokuphuphuthekile i-AI noma ukuyisebenzisa kabi kungaletha ubungozi obusha. Ngezansi, sibeka izinto ezikhathazayo ezinkulu nezihibe, kanye nomongo wazo ngamunye:

Ukusetshenziswa Okumelene Nezigebengu Ze-Cybercriminal: Amandla afanayo okukhiqiza asiza abavikeli ukuthi banike amandla abahlaseli. Abalingisi abasabisayo sebevele basebenzisa i-AI ekhiqizayo ukwenza ama-imeyili obugebengu bokweba imininingwane ebucayi, bakhe abantu mbumbulu namavidiyo mbumbulu wobunjiniyela bezenhlalo, ukuthuthukisa uhlelo olungayilungele ikhompuyutha oluhlala lushintsha ukuze lugweme ukutholwa, ngisho nezici zokugebenga ngokuzenzakalela ( Iyini I-Generative AI ku-Cybersecurity? - Palo Alto Networks ). Cishe ingxenye (46%) yabaholi be-cybersecurity bakhathazekile ngokuthi i-AI ekhiqizayo izoholela ekuhlaselweni kwezitha okuthuthuke kakhulu ( Ukuvikeleka kwe-Generative AI: Amathrendi, Izinsongo & Namasu Okunciphisa ). Lo “mjaho wezikhali we-AI” usho ukuthi njengoba abavikeli besebenzisa i-AI, abahlaseli ngeke basalele emuva (empeleni, bangase babe phambili kwezinye izindawo, besebenzisa amathuluzi e-AI angalawulwa). Izinhlangano kufanele zilungele izinsongo ezithuthukisiwe ze-AI ezivame kakhulu, eziyinkimbinkimbi, futhi okunzima ukuzilandelela.
I-AI Halucinations kanye Nokunemba: Amamodeli e-Generative AI angakhiqiza imiphumela ezwakalayo kodwa engalungile noma edukisayo - into eyaziwa ngokuthi i-allucination. Kumongo wokuphepha, i-AI ingase ihlaziye isigameko futhi iphethe ngephutha ukuthi ubungozi obuthile bekuyimbangela, noma ingase ikhiqize iskripthi sokulungisa esinephutha esihluleka ukuqukatha ukuhlasela. Lawa maphutha angaba yingozi uma ethathwa ngokusobala. Njengoba Idatha ye-NTT ixwayisa, “i-AI ekhiqizayo ingase ikhiphe okuqukethwe okungelona iqiniso, futhi lesi simo sibizwa ngokuthi ama-allucinations… okwamanje kunzima ukuwaqeda ngokuphelele” ( Izingozi Zokuphepha Ze-Generative AI kanye Nezinyathelo Zokulwa, kanye Nomphumela Wayo Ku-Cybersecurity | Iqembu le-NTT DATA ). Ukuthembela ngokweqile ku-AI ngaphandle kokuqinisekisa kungaholela emizamweni engaqondiwe kahle noma umuzwa wokuphepha ongamanga. Isibonelo, i-AI ingase ihlabe umkhosi ngamanga isistimu ebucayi njengephephile uma ingekho, noma ngokuphambene, icuphe ukwethuka “ngokubona” ukuphulwa okungakaze kwenzeke. Ukuqinisekiswa okuqinile kwemiphumela ye-AI kanye nokuba nabantu ku-loop ezinqumweni ezibalulekile kubalulekile ukuze kuncishiswe lobu bungozi.
Okuhle Okungamanga Nokubi: Okuhlobene nokubona izinto ezingekho, uma imodeli ye-AI ingaqeqeshwanga kahle noma ilungiselelwe kahle, ingase ibike ngokweqile umsebenzi ongalungile njengononya (okukhona okungelona iqiniso) noma, okubi nakakhulu, igeje izinsongo zangempela (okungalungile okungamanga) ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity ). Izexwayiso ezingamanga ezeqile zingasibekela amaqembu okuvikela futhi ziholele ekukhathaleni okuqwashile (ukuhlehlisa izinzuzo ezisebenza kahle kakhulu ezithenjisiwe yi-AI), kuyilapho ukutholwa okugejiwe kushiya inhlangano idaluliwe. Ukushuna amamodeli akhiqizayo ukuze uthole ibhalansi efanele kuyinselele. Indawo ngayinye ihlukile, futhi i-AI ingase ingasebenzi kahle ngaphandle kwebhokisi. Ukufunda okuqhubekayo kuyinkemba esika nhlangothi zombili futhi - uma i-AI ifunda empendulweni ehlanekezelwe noma endaweni eshintshayo, ukunemba kwayo kungashintshashintsha. Amathimba okuvikela kumele aqaphe ukusebenza kwe-AI futhi alungise imikhawulo noma anikeze impendulo yokulungisa kumamodeli. Ezimweni eziphakeme kakhulu (njengokutholwa kokungena kwengqalasizinda ebalulekile), kungase kube ubuhlakani ukusebenzisa iziphakamiso ze-AI ngokuhambisana namasistimu akhona isikhathi esithile, ukuze kuqinisekiswe ukuthi ayahambisana futhi ayahambisana kunokungqubuzana.
Ubumfihlo Bedatha Nokuvuza: Amasistimu e-AI akhiqizayo ngokuvamile adinga inani elikhulu ledatha yokuqeqeshwa nokusebenza. Uma lawa mamodeli asuselwe kumafu noma engavaliwe ngendlela efanele, kunobungozi bokuthi ulwazi olubucayi lungavuza. Abasebenzisi bangase baphakele idatha yokuphathelene noma idatha yomuntu siqu kusevisi ye-AI bengaqondile (bacabange ukucela i-ChatGPT ukuthi ifinyeze umbiko wesigameko esiyimfihlo), futhi leyo datha ingaba yingxenye yolwazi lwemodeli. Ngempela, ucwaningo lwakamuva luthole ukuthi i-55% yokokufaka kumathuluzi e-AI akhiqizayo aqukethe ulwazi olubucayi noma oluhlonza umuntu siqu , okuphakamisa ukukhathazeka okujulile mayelana nokuvuza kwedatha ( Ukuphepha kwe-AI Okukhiqizayo: Amathrendi, Izinsongo & Namasu Okunciphisa ). Ukwengeza, uma i-AI iqeqeshwe ngedatha yangaphakathi futhi ibuzwa ngezindlela ezithile, ingase ikhiphe izingcezu zaleyo datha ebucayi komunye umuntu. Izinhlangano kufanele zisebenzisa izinqubomgomo zokuphatha idatha eziqinile (isb. ukusebenzisa i-AI yasendaweni noma izimo ze-AI eziyimfihlo ezintweni ezibucayi) futhi zifundise abasebenzi mayelana nokunganamathiseli ulwazi oluyimfihlo kumathuluzi e-AI omphakathi. Imithetho yobumfihlo (GDPR, njll.) nayo iyasebenza - ukusebenzisa idatha yomuntu siqu ukuqeqesha i-AI ngaphandle kwemvume efanele noma ukuvikelwa kungase kungqubuzane nemithetho.
Ukuphepha Kwemodeli Nokukhohlisa: Amamodeli e-Generative AI ngokwawo angaba okuqondiwe. Izitha zingase zizame ubuthi bemodeli , ukuphakela idatha enonya noma edukisayo phakathi nesigaba sokuqeqesha noma sokuqeqesha kabusha ukuze i-AI ifunde amaphethini angalungile ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity ). Isibonelo, umhlaseli angase abe nobuthi obuyingozi kudatha ye-intel ukuze i-AI yehluleke ukubona uhlelo olungayilungele ikhompuyutha lomhlaseli njengolunonya. Elinye iqhinga elokujova ngokushesha noma ukukhohlisa kokukhiphayo , lapho umhlaseli ethola indlela yokukhipha okokufaka ku-AI okubangela ukuba iziphathe ngendlela engahlosiwe - mhlawumbe ingaziba imingcele yayo yokuphepha noma ukuveza ulwazi okungafanele iluveze (njengemiyalo yangaphakathi noma idatha). Ukwengeza, kunobungozi bokubalekela imodeli : abahlaseli benza okokufaka okuklanyelwe ukukhohlisa i-AI. Lokhu sikubona ezibonelweni eziphikisanayo - idatha ephazamisekile kancane umuntu ayibona njengejwayelekile kodwa i-AI ibeka ngokungeyikho. Ukuqinisekisa ukuthi uchungechunge lokuhlinzeka nge-AI luvikelekile (ubuqotho bedatha, ukulawula ukufinyelela kwemodeli, ukuhlolwa kokuqina komphikisi) yingxenye entsha kodwa edingekayo ye-cybersecurity lapho kusetshenziswa lawa mathuluzi ( Iyini I-Generative AI ku-Cybersecurity? - I-Palo Alto Networks ).
Ukwethembela Ngokweqile kanye Nokuguguleka Kwamakhono: Kunobungozi obuncane bokuthi izinhlangano zingathembela ngokweqile ku-AI futhi zivumele amakhono omuntu ukuthi afane. Uma abahlaziyi abancane bethemba imiphumela ye-AI ngokungaboni, bangase bangathuthukisi ukucabanga okujulile kanye nomuzwa odingekayo lapho i-AI ingatholakali noma ingalungile. Isimo okufanele sigwenywe ithimba lokuvikela elinamathuluzi amahle kodwa elingazi ukuthi lisebenza kanjani uma lawo mathuluzi ehla (okufana nabashayeli bezindiza abathembele ngokweqile ekushayeleni okuzenzakalelayo). Ukuzivocavoca okuvamile ngaphandle kosizo lwe-AI nokukhuthaza umqondo wokuthi i-AI ingumsizi, hhayi i-oracle engenaphutha, kubalulekile ukugcina abahlaziyi abangabantu bebukhali. Abantu kumele bahlale bengabenzi bezinqumo bokugcina, ikakhulukazi ezinqumweni ezinomphumela omkhulu.
Izinselelo Zokuziphatha Nokuhambisana: Ukusetshenziswa kwe-AI ekuvikelekeni ku-inthanethi kuphakamisa imibuzo yokuziphatha futhi kungase kubangele izinkinga zokuthobela imithetho. Isibonelo, uma uhlelo lwe-AI lufaka isisebenzi ngokungalungile njengomuntu ongaphakathi ononya ngenxa yesimo esingaqondakali, lungalimaza ngokungafanele isithunzi noma umsebenzi walowo muntu. Izinqumo ezenziwe yi-AI zingase zibe opaque (inkinga "yebhokisi elimnyama"), okwenza kube nzima ukuchaza abacwaningi mabhuku noma abalawuli ukuthi kungani izenzo ezithile zenziwe. Njengoba okuqukethwe okukhiqizwe yi-AI kuya kwanda kakhulu, ukuqinisekisa ukuthi izinto obala nokugcina ukuziphendulela kubalulekile. Abalawuli sebeqalile ukuhlolisisa i-AI – uMthetho we-AI we-EU, isibonelo, uzobeka izidingo ezinhlelweni ze-AI “eziyingozi kakhulu”, futhi i-cybersecurity AI ingase iwele kuleso sigaba. Izinkampani zizodinga ukuzulazula kule mithetho futhi ngokunokwenzeka zibambelele kumazinga afana ne-NIST AI Risk Management Framework ukuze zisebenzise i-AI ekhiqizayo ngokuzibophezela ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Ukuthobela kudlulela nasekunikezeni amalayisense futhi: ukusebenzisa imithombo evulekile noma amamodeli ezinkampani zangaphandle kungase kube nemigomo ekhawulela ukusetshenziswa okuthile noma kudinga ukuthuthukiswa kokwabelana.

Kafushane, i-AI ekhiqizayo ayiyona inhlamvu yesiliva - uma ingenziwanga ngokucophelela, ingethula ubuthakathaka obusha njengoba ixazulula abanye. Ucwaningo lwe-World Economic Forum lwango-2024 lwagqamisa ukuthi cishe ama-47% ezinhlangano acaphuna intuthuko ku-AI ekhiqizwayo ngabahlaseli njengento ekhathaza kakhulu, okuyenza “umthelela ophathelene kakhulu ne-generative AI” ku-cybersecurity ( [PDF] Global Cybersecurity Outlook 2025 | World Economic Forum ) ( I-Generative AI ku-Cybersecursive LLM Review ). Ngakho-ke, izinhlangano kufanele zisebenzisa indlela elinganiselayo: zisebenzise izinzuzo ze-AI kuyilapho zilawula ngokuqinile lezi zingozi ngokubusa, ukuhlola, nokwengamela komuntu. Ngokulandelayo sizoxoxa ngendlela yokufinyelela lokho kulinganisela.

I-future Outlook: Indima Eguqukayo Ye-AI Ye-Cybersecurity

Uma sibheka phambili, i-AI ekhiqizayo isilungele ukuba yingxenye ebalulekile yesu lokuvikeleka ku-inthanethi - futhi ngokufanayo, ithuluzi izitha ze-inthanethi ezizoqhubeka nokulisebenzisa. I- cat-and-mouse eguqukayo izosheshisa, nge-AI ezinhlangothini zombili zocingo. Nansi eminye imininingwane ebheke phambili yokuthi i-AI ekhiqizayo ingakumisa kanjani ukuphepha ku-inthanethi eminyakeni ezayo:

I-AI-Augmented Cyber Defense Iba Okujwayelekile: Ngo-2025 nangale kwalokho, singalindela ukuthi izinhlangano eziningi eziphakathi nendawo ezinkulu zizobe sezihlanganise amathuluzi aqhutshwa yi-AI emisebenzini yazo yezokuphepha. Njengoba i-antivirus kanye nezindonga zomlilo zijwayelekile namuhla, amakhophi e-AI kanye nezinhlelo zokuthola okungaqondakali zingase zibe izingxenye eziyisisekelo zezakhiwo zokuphepha. Lawa mathuluzi cishe azogxila kakhulu - ngokwesibonelo, amamodeli e-AI ahlukile alungiselwe ukuphepha kwamafu, ukuqapha idivayisi ye-IoT, ukuphepha kwekhodi yohlelo lokusebenza, njalo njalo, wonke asebenza ekhonsathini. Njengoba esinye isibikezelo siphawula, "ku-2025, i-AI ekhiqizayo izoba yingxenye ye-cybersecurity, ivumela izinhlangano ukuthi zivikele ezinsongweni eziyinkimbinkimbi neziguqukayo ngokuqhubekayo" ( Ingasetshenziswa Kanjani I-Generative AI ku-Cybersecurity ). I-AI izothuthukisa ukutholwa kosongo lwesikhathi sangempela, yenze ngokuzenzakalelayo izenzo eziningi zokuphendula, futhi isize amaqembu okuvikela aphathe imiqulu yedatha emikhulu kakhulu kunaleyo abengayenza mathupha.
Ukufunda Okuqhubekayo Nokujwayela: Izinhlelo ze-AI ezikhiqizayo zesikhathi esizayo ku-cyber zizoba ngcono ekufundeni ngokundiza kusuka ezehlakalweni ezintsha kanye nobuhlakani obusongelayo, zibuyekeze isisekelo sazo solwazi esikhathini esiseduze-sangempela. Lokhu kungase kuholele ekuvikelweni okuguquguqukayo ngempela - cabanga nge-AI efunda ngomkhankaso omusha wobugebengu bokweba imininingwane ebucayi oshaya enye inkampani ekuseni futhi ntambama isivele izilungisile izihlungi ze-imeyili zenkampani yakho ukuze iphendule. Izinsizakalo zokuphepha ze-AI ezisekelwe emafini zingase zisize lolu hlobo lokufunda okuhlangene, lapho imininingwane engaziwa evela enhlanganweni eyodwa izuzisa bonke ababhalisile (okufana nokusongela ukwabelana kwe-intel, kodwa okuzenzakalelayo). Kodwa-ke, lokhu kuzodinga ukuphathwa ngokucophelela ukuze kugwenywe ukwabelana ngolwazi olubucayi nokuvimbela abahlaseli ekufakeni idatha embi kumamodeli abiwe.
Ukuhlangana kwe-AI kanye ne-Cybersecurity Talent: Isethi yamakhono ochwepheshe be-cybersecurity izovela ukuze ifake ubuchwepheshe ku-AI nesayensi yedatha. Njengoba nje abahlaziyi banamuhla befunda izilimi zemibuzo nokubhala, abahlaziyi bakusasa bangase bahlele kahle amamodeli e-AI noma babhale “izincwadi zokudlala” ukuze i-AI iyisebenzise. Singase sibone izindima ezintsha njengokuthi “AI Security Trainer” noma “Cybersecurity AI Engineer” - abantu abasebenza ngokukhethekile ukulungisa amathuluzi e-AI ukuze ahambisane nezidingo zenhlangano, baqinisekise ukusebenza kwabo, kanye nokuqinisekisa ukuthi basebenza ngokuphephile. Ngakolunye uhlangothi, ukucatshangelwa kwe-cybersecurity kuzokhula kube nomthelela ekuthuthukisweni kwe-AI. Amasistimu e-AI azokwakhiwa ngezici zokuphepha kusukela phansi (izakhiwo ezivikelekile, ukutholwa kokuphazamiseka, amalogi okucwaninga ezinqumo ze-AI, njll.), kanye nezinhlaka ze -AI ezinokwethenjelwa (ezinobulungiswa, ezichazekayo, eziqinile, nezivikelekile) zizoqondisa ukusetshenziswa kwazo ezimweni ezibucayi zokuphepha.
Ukuhlasela Okunamandla Kwe-AI Okuthuthuke Kakhudlwana: Ngeshwa, indawo esongelayo izophinde iguquke nge-AI. Silindele ukusetshenziswa kaningi kwe-AI ukuze sithole ubungozi bosuku oluyiziro, ukwenza ubugebengu bokweba imininingwane ebucayi obuqondiswe kakhulu (isb. I-AI iklwebha inkundla yezokuxhumana ukuze idale isiyengo esiklanywe kahle), futhi sikhiqize amazwi noma amavidiyo ajulile akhohlisayo ukuze adlule ukuqinisekiswa kwebhayomethrikhi noma ukukhwabanisa. Ama-ejenti okugebenga azenzakalelayo angase avele akwazi ukuhlasela ngokuzimela izigaba eziningi (ukuhlola, ukuxhashazwa, ukunyakaza okuhlangene, njll.) ngokugada komuntu okuncane. Lokhu kuzocindezela abavikeli ukuthi bathembele ku-AI - empeleni i-automation vs. automation . Okunye ukuhlaselwa kungenzeka ngesivinini somshini, njengamabhothi e-AI azama izimvume ze-imeyili zobugebengu bokweba imininingwane ebucayi eyinkulungwane ukubona ukuthi yimuphi othola izihlungi ezedlule. Izivikelo ze-Cyber zizodinga ukusebenza ngesivinini esifanayo nokuvumelana nezimo ukuze ziqhubeke ( Iyini i-Generative AI ku-Cybersecurity? - I-Palo Alto Networks ).
I-Regulation kanye ne-Ethical AI Ekuvikelekeni: Njengoba i-AI igxila ngokujulile emisebenzini ye-cybersecurity, kuzoba nokucutshungulwa okukhulu kanye nomthetho ngokunokwenzeka ukuze kuqinisekiswe ukuthi lezi zinhlelo ze-AI zisetshenziswa ngokuzibophezela. Singalindela izinhlaka namazinga aqondene ne-AI ekuvikelekeni. Ohulumeni bangase babeke imihlahlandlela yokubeka izinto obala - isb, ezidinga ukuthi izinqumo ezibalulekile zokuphepha (njengokunqamula ukufinyelela kwesisebenzi emisebenzini esolwayo enonya) ngeke zenziwe i-AI iyodwa ngaphandle kokubuyekezwa komuntu. Kungase futhi kube nezitifiketi zemikhiqizo yezokuphepha ye-AI, ukuze kuqinisekiswe abathengi ukuthi i-AI ihlolelwe ukuchema, ukuqina, nokuphepha. Ngaphezu kwalokho, ukubambisana kwamazwe ngamazwe kungase kukhule mayelana nezinsongo ze-cyber ezihlobene ne-AI; isibonelo, izivumelwano zokuphatha i-disinformation edalwe yi-AI noma izinkambiso eziphikisana nezikhali ezithile ze-inthanethi eziqhutshwa yi-AI.
Ukuhlanganiswa ne-Broader AI kanye ne-IT Ecosystems: I-Generative AI ku-cybersecurity cishe izohlanganisa nezinye izinhlelo ze-AI namathuluzi okuphatha i-IT. Isibonelo, i-AI ephethe ukuthuthukiswa kwenethiwekhi ingasebenza ne-AI yokuphepha ukuze kuqinisekiswe ukuthi izinguquko azivuli izintuba. Izibalo zebhizinisi eziqhutshwa yi-AI zingase zabelane ngedatha nama-AI okuphepha ukuze kuxhunywe okudidayo (njengokwehla kokuthengisa okungazelelwe nenkinga yewebhusayithi engenzeka ngenxa yokuhlaselwa). Empeleni, i-AI ngeke ihlale esiseleni - izoba yingxenye yendwangu enkulu ehlakaniphile yokusebenza kwenhlangano. Lokhu kuvula amathuba okulawulwa kobungozi okuphelele lapho idatha yokusebenza, idatha yosongo, ngisho nedatha yokuphepha ebonakalayo ingahlanganiswa yi-AI ukunikeza umbono we-360-degree wokuma kwenhlangano kwezokuphepha.

Ngokuhamba kwesikhathi, ithemba liwukuthi i-AI ekhiqizayo izosiza ukutshekisa ibhalansi ivune abavikeli. Ngokuphatha isikali nobunkimbinkimbi bezindawo zesimanje ze-IT, i-AI ingenza i-cyberspace ivikeleke kakhulu. Nokho, uhambo, futhi kuzoba nezinhlungu ezikhulayo njengoba sicwenga lobu buchwepheshe futhi sifunda ukubethemba ngendlela efanele. Izinhlangano ezihlala zinolwazi futhi ezitshala imali ekutholweni kwe-AI okunesibopho kwezokuphepha kungenzeka kube yizona ezisesimweni esihle sokuzulazula ezinsongweni zesikhathi esizayo.

Njengoba umbiko wakamuva we-cybersecurity ka-Gartner uphawula, "ukuvela kwamacala okusebenzisa i-AI akhiqizayo (kanye nezingozi) kudala ingcindezi yokuguqulwa" ( Amathrendi e-Cybersecurity: Resilience Through Transformation - Gartner ). Labo abavumelana nezimo bazosebenzisa i-AI njengozakwethu onamandla; labo abasalele bangazithola bedlulwa yizitha ezinikwe amandla yi-AI. Iminyaka embalwa ezayo kuzoba isikhathi esibalulekile ekuchazeni ukuthi i-AI iyilungisa kanjani kabusha inkundla yempi ye-cyber.

Izinyathelo Ezisebenzayo Zokuthola I-Generative AI ku-Cybersecurity

Emabhizinisini ahlola ukuthi angayisebenzisa kanjani i-AI ekhiqizayo esu lawo lokuphepha ku-inthanethi, nazi ezinye izinto ezisebenzayo nezincomo zokuqondisa ukutholwa okunesibopho nokuphumelelayo:

Qala Ngezemfundo Nokuqeqeshwa: Qinisekisa ukuthi ithimba lakho lezokuphepha (kanye nabasebenzi be-IT ababanzi) bayaqonda ukuthi yini i-AI ekhiqizayo engakwazi ukuyenza futhi engakwazi ukuyenza. Nikeza ngoqeqesho ngezisekelo zamathuluzi okuvikela ashayelwa yi-AI futhi ubuyekeze izinhlelo zakho zokuqwashisa ngezokuphepha kubo bonke abasebenzi ukuze bavikele izinsongo ezinikwe amandla i-AI. Isibonelo, fundisa abasebenzi ukuthi i-AI ingakhiqiza kanjani imikhonyovu yobugebengu bokweba imininingwane ebucayi nezingcingo ezijulile. Ngesikhathi esifanayo, qeqesha abasebenzi ngokusetshenziswa okuphephile nokugunyaziwe kwamathuluzi e-AI emsebenzini wabo. Abasebenzisi abanolwazi mancane amathuba okuthi baphathe kabi i-AI noma babe yizisulu zokuhlaselwa okuthuthukisiwe kwe-AI ( Ingasetshenziswa Kanjani I-AI Ekhiqizayo Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ).
Chaza Izinqubomgomo Zokusetshenziswa Ze-AI Ecacile: Phatha i-AI ekhiqizayo njenganoma ibuphi ubuchwepheshe obunamandla - ngokubusa. Thuthukisa izinqubomgomo ezicacisa ukuthi ubani ongasebenzisa amathuluzi e-AI, imaphi amathuluzi agunyaziwe, nokuthi ngaziphi izinjongo. Faka phakathi imihlahlandlela yokuphatha idatha ebucayi (isb. ukungaphakelwa kwedatha eyimfihlo kumasevisi e-AI angaphandle) ukuze kuvinjelwe ukuvuza. Njengesibonelo, ungase uvumele amalungu ethimba lezokuphepha kuphela ukuthi asebenzise umsizi we-AI wangaphakathi ukuze aphendule ngesigameko, futhi ukumaketha kungasebenzisa i-AI ehloliwe kokuqukethwe - wonke umuntu ukhawulelwe. Manje izinhlangano eziningi zikhuluma ngokusobala ekhiqizayo i-AI kuzinqubomgomo zazo ze-IT, futhi izinhlangano eziholayo ezisezingeni zikhuthaza izinqubomgomo zokusetshenziswa okuphephile kunokuvinjelwa okuqondile ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Qiniseka ukuthi ukhuluma ngale mithetho kanye nesizathu salokho kubo bonke abasebenzi.
Nciphisa “Ithunzi I-AI” kanye Nokuqapha Ukusetshenziswa: Ngokufana nesithunzi se-IT, “ithunzi AI” livela lapho abasebenzi beqala ukusebenzisa amathuluzi noma amasevisi e-AI ngaphandle kolwazi lwe-IT (isb. umthuthukisi osebenzisa umsizi wekhodi ye-AI ongagunyaziwe). Lokhu kungethula izingozi ezingabonakali. Sebenzisa izinyathelo zokuthola nokulawula ukusetshenziswa kwe-AI okungagunyaziwe . Ukuqapha inethiwekhi kungahlaba umkhosi uxhumo kuma-AI API adumile, futhi izinhlolovo noma ukuhlola kwamathuluzi kungadalula ukuthi abasebenzi basebenzisa ini. Nikeza ezinye izindlela ezigunyaziwe ukuze izisebenzi ezinezisusa ezinhle zingalingeki ukuba ziziphathe kabi (ngokwesibonelo, hlinzeka nge-akhawunti esemthethweni ye-ChatGPT Enterprise uma abantu beyithola iwusizo). Ngokuletha ukusetshenziswa kwe-AI ekukhanyeni, amaqembu ezokuphepha angahlola futhi alawule ubungozi. Ukuqapha nakho kuyisihluthulelo - ngena imisebenzi yethuluzi le-AI kanye nemiphumela ngangokunokwenzeka, ngakho-ke kukhona umkhondo wokucwaninga wezinqumo ezithonywe yi-AI ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ).
Sebenzisa i-AI Ngokuzivikela - Ungawi Ngemuva: Yazi ukuthi abahlaseli bazosebenzisa i-AI, ngakho-ke ukuzivikela kwakho kufanele nakho. Khomba izindawo ezimbalwa ezinomthelela omkhulu lapho i-AI ekhiqizayo ingasiza khona ngokushesha imisebenzi yakho yezokuphepha (mhlawumbe isivivinyo sokuxwayisa, noma ukuhlaziywa kwelogi okuzenzakalelayo) futhi iqhube amaphrojekthi okuhlola. Khulisa ukuzivikela kwakho ngesivinini nesikali se-AI ukuze umelane nezinsongo ezihamba ngokushesha ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? Izibonelo Ezingu-10 Zangempela Zomhlaba ). Ngisho nokuhlanganiswa okulula, njengokusebenzisa i-AI ukufingqa imibiko yohlelo olungayilungele ikhompuyutha noma ukukhiqiza imibuzo yokuzingela okusongelayo, kungasindisa amahora abahlaziyi. Qala kancane, hlola imiphumela, futhi uphindaphinde. Impumelelo izokwakha icala lokwamukelwa okubanzi kwe-AI. Umgomo uwukusebenzisa i-AI njengesiphindaphindisi esinamandla - isibonelo, uma ukuhlaselwa kobugebengu bokweba imininingwane ebucayi kungaphezu kwedeskithophu yakho yosizo, sebenzisa isihlungi se-imeyili se-AI ukuze wehlise leyo volumu ngokuqhubekayo.
Tshala Ngemikhuba Ye-AI Evikelekile Nelungile: Lapho usebenzisa i-AI ekhiqizayo, landela intuthuko evikelekile nezinqubo zokuphakela. Sebenzisa amamodeli ayimfihlo noma azibambele wona wemisebenzi ebucayi ukuze ugcine ukulawula idatha. Uma usebenzisa amasevisi ezinkampani zangaphandle ze-AI, buyekeza izindlela zabo zokuphepha nezobumfihlo (ukubethela, izinqubomgomo zokugcinwa kwedatha, njll.). Faka izinhlaka ze-AI zokulawula ubungozi (njenge-NIST's AI Risk Management Framework noma isiqondiso se-ISO/IEC) ukuze kubhekwane nezinto ezifana nokuchema, ukuchaza, nokuqina kumathuluzi akho e-AI ( Ingasetshenziswa Kanjani I-Generative AI Ku-Cybersecurity? 10 Izibonelo Zomhlaba Wangempela ). Futhi hlela ukubuyekezwa kwamamodeli/amapeshi njengengxenye yokulungiswa - amamodeli e-AI angaba "nobungozi" nawo (isb. angase adinge ukuqeqeshwa kabusha uma eqala ukukhukhuleka noma uma kutholwa uhlobo olusha lokuhlasela oluphikisana nemodeli). Ngokubhaka ukuphepha nezimiso zokuziphatha ekusetshenzisweni kwakho kwe-AI, wakha ukwethemba imiphumela futhi uqinisekisa ukuthotshelwa kwemithetho evelayo.
Gcina Abantu Ku-Loop: Sebenzisa i-AI ukusiza, hhayi ukufaka esikhundleni ngokuphelele, ukwahlulela komuntu ku-cybersecurity. Nquma amaphuzu ezinqumo lapho kudingeka khona ukuqinisekiswa komuntu (ngokwesibonelo, i-AI ingase ibhale umbiko wesigameko, kodwa umhlaziyi uyawubuyekeza ngaphambi kokusatshalaliswa; noma i-AI ingase iphakamise ukuvinjwa kwe-akhawunti yomsebenzisi, kodwa umuntu uyasigunyaza leso senzo). Lokhu akuvimbi kuphela amaphutha e-AI ukuthi angahloliwe, kodwa futhi kusiza iqembu lakho ukuthi lifunde ku-AI nangokuphambene. Khuthaza ukuhamba komsebenzi okuhlanganyelwe: abahlaziyi kufanele bazizwe bekhululekile ukubuza imiphumela ye-AI kanye nokuhlola ukuhlanzeka kwengqondo. Ngokuhamba kwesikhathi, le ngxoxo ingathuthukisa kokubili i-AI (ngempendulo) kanye namakhono abahlaziyi. Empeleni, klama izinqubo zakho ngendlela yokuthi i-AI namandla omuntu aphelelisane - i-AI iphatha ivolumu nesivinini, abantu basingatha ukungaqondakali nezinqumo zokugcina.
Linganisa, Gada, futhi Ulungise: Okokugcina, phatha amathuluzi akho e-AI akhiqizayo njengezingxenye eziphilayo ze-ecosystem yakho yezokuphepha. Ngokuqhubekayo kala ukusebenza kwabo - ingabe banciphisa izikhathi zokuphendula isigameko? Ukubamba izinsongo ngaphambi kwesikhathi? Linjani izinga lephozithivu elingamanga lithrendayo? Cela impendulo eqenjini: ingabe izincomo ze-AI ziwusizo, noma idala umsindo? Sebenzisa lawa mamethrikhi ukuze wenze ngcono amamodeli, ubuyekeze idatha yokuqeqeshwa, noma ulungise indlela i-AI ehlanganiswe ngayo. Izinsongo ze-Cyber nezidingo zebhizinisi ziyashintsha, futhi amamodeli akho e-AI kufanele abuyekezwe noma aqeqeshwe kabusha ngezikhathi ezithile ukuze ahlale esebenza kahle. Yiba nohlelo lokuphatha okuyimodeli, okuhlanganisa nokuthi ubani onesibopho sokuyigcina isesimweni esihle nokuthi ibuyekezwa kangaki. Ngokuphatha umjikelezo wokuphila we-AI ngenkuthalo, uyaqinisekisa ukuthi ihlala iyimpahla, hhayi isikweletu.

Sengiphetha, i-AI ekhiqizayo ingathuthukisa kakhulu amakhono e-cybersecurity, kodwa ukwamukelwa ngempumelelo kudinga ukuhlela okucatshangelwayo kanye nokwengamela okuqhubekayo. Amabhizinisi afundisa abantu bawo, abeka iziqondiso ezicacile, futhi ahlanganise i-AI ngendlela elinganiselayo, evikelekile azovuna imivuzo yokuphatha izinsongo ezisheshayo, ezihlakaniphile. Lezo zindlela ezithathwayo zinikeza umhlahlandlela: hlanganisa ubuchwepheshe bomuntu ne-AI automation, imboze izisekelo zokubusa, futhi igcine ubungqabavu njengoba kokubili ubuchwepheshe be-AI kanye nesimo esisongelayo kuguquka ngokungenakugwenywa.

Ngokuthatha lezi zinyathelo ezisebenzayo, izinhlangano zingaphendula ngokuzethemba umbuzo othi “Ingasetshenziswa kanjani i-AI ekhiqizayo ekuvikelekeni kwe-inthanethi?” - hhayi nje ngethiyori, kodwa ekusebenzeni kwansuku zonke - futhi ngaleyo ndlela baqinise ukuzivikela kwabo emhlabeni wethu owandayo wedijithali kanye ne-AI. ( Ingasetshenziswa Kanjani I-Generative AI ku-Cybersecurity )

Amaphepha amhlophe ongase uthande ukuwafunda ngemva kwaleli:

🔗 Imisebenzi I-AI Engakwazi Ukuyishintsha Futhi Imiphi Imisebenzi I-AI Ezoyishintsha?
Hlola umbono wembulunga yonke wokuthi yiziphi izindima eziphephile ku-automation nokuthi yiziphi ezingaphephile.

🔗 Ingabe i-AI Ingakwazi Ukubikezela Imakethe Yezitoko?
Ukubhekisisa imikhawulo, impumelelo, kanye nezinganekwane ezizungeze ikhono le-AI lokubikezela umnyakazo wemakethe.

🔗 Yini Okungathenjelwa Kuyo I-Generative AI Ngaphandle Kokungenelela Komuntu?
Qonda lapho i-AI ingasebenza khona ngokuzimela nalapho ukuqapha komuntu kusabalulekile.

Buyela kubhulogi

Izwe/isifunda