Abadlali abasabisayo abagibele ku- ukuduma kwe-ChatGPT sethule elinye futhi ithuluzi lokugebenga i-copycat elinikezela ngezinsizakalo ezifanayo ze-charbot kuhlelo lokusebenza olukhiqizayo lwangempela olusekelwe ku-AI kodwa elihloselwe ukukhuthaza umsebenzi onobungozi.

Abacwaningi bathole izikhangiso ezifakwe kuWebhu Emnyama zethuluzi lokugebenga eliqhutshwa yi-AI elibizwa ngokuthi “FraudGPT,” elithengiswa ngokubhaliselwe futhi belizungeza kuTelegram kusukela ngoMgqibelo, abacwaningi baseNetenrich baveze. eposini eshicilelwe ngoJulayi 25. I-FraudGPT iqala ku-$200 ngenyanga futhi ikhuphukela ku-$1,700 ngonyaka, futhi ihloselwe ukusiza abagebengu basebenzise ibhizinisi labo elibi ngosizo lwe-AI. Umlingisi uthi unokuthengisa okungaphezulu kwe-3,000 okuqinisekisiwe nokubuyekezwa kuze kube manje ku-FraudGPT.

Elinye ithuluzi elifanayo, eliqhutshwa yi-AI, I-WormGPT, ibilokhu isakazwa kusukela ngoJulayi 13, futhi yachazwa ngokuningiliziwe embikweni we-SlashNext. Njenge-ChatGPT, lezi ezivelayo I-AI ephikisanayo amathuluzi nawo asekelwe kumamodeli aqeqeshwe emithonjeni emikhulu yedatha, futhi angakhiqiza umbhalo ofana nomuntu ngokusekelwe kokokufaka abakutholayo.

Amathuluzi "abonakala ephakathi kokuthambekela kokuqala kokuthi abalingisi abasabisayo bakhe izici ze-AI ezikhiqizayo kumathuluzi abo," uJohn Bambenek, umzingeli oyinhloko osongela eNetenrich, utshela i-Dark Reading. "Ngaphambi kwalokhu, ukuxoxisana kwethu mayelana nesimo esisongelayo bekungokwengqondo."

I-FraudGPT - ezikhangisweni ezikhangisiwe "njenge-bot ngaphandle kwemikhawulo, imithetho, [kanye] nemingcele" - ithengiswa ngumdlali osongelayo othi ungumdayisi oqinisekisiwe ezimakethe ezihlukahlukene ezingaphansi komhlaba ezimnyama zeWebhu, okuhlanganisa i-Empire, WHM, Torrez, World , AlphaBay, and Versus.

AmaCybercriminals Manje Ahlome Nge-AI Chatbots

Kokubili i-WormGPT ne-FraudGPT ingasiza abahlaseli ukuthi basebenzise i-AI ukuze bazuze lapho benza imikhankaso yobugebengu bokweba imininingwane ebucayi, bekhiqiza imilayezo ehloselwe ukucindezela izisulu ekuweleni ekutholweni kwe-imeyili yebhizinisi (BEC), neminye imikhonyovu esekelwe ku-imeyili, yabaqalayo.

I-FraudGPT futhi ingasiza abalingisi abasabisayo benze inqwaba yezinye izinto ezimbi, njengokuthi: ukubhala ikhodi enonya; ukudala i-malware engabonakali; ukuthola imigqomo engeyona eye-VBV; ukudala amakhasi obugebengu bokweba imininingwane ebucayi; amathuluzi okugebenga isakhiwo; ukuthola amaqembu okugebenga, amasayithi, nezimakethe; ukubhala amakhasi nezinhlamvu zomkhonyovu; ukuthola ukuvuza kanye nokuba sengozini; nokufunda ukwenza ikhodi noma ukugebenga. 

Noma kunjalo, kubonakala sengathi ukusiza abahlaseli bakhe imikhankaso yobugebengu bokweba imininingwane ebucayi kusengenye yezindlela eziyinhloko zokusetshenziswa kwethuluzi elifana ne-FraudGPT, ngokusho kwe-Netenrich. Ubuchule baleli thuluzi kulokhu buye baphakanyiswa nasemibhalweni yokuphromotha evela kuWebhu Emnyama ebonisa ukuthi i-FraudGPT ingakhiqiza kanjani i-imeyili esalungiswa “ezoyenga abamukeli ukuthi bachofoze isixhumanisi esinonya esihlinzekiwe,” kusho u-Krishnan.

I-Jailbreaking ChatGPT's Ethical Guardrails

Ngenkathi I-ChatGPT nayo ingaxhashazwa njengethuluzi le-Hacker ukubhala ama-imeyili akhiwe emphakathini, kunezivikelo zesimilo ezikhawulela lokhu kusetshenziswa. Kodwa-ke, ukwanda okukhulayo kwamathuluzi aqhutshwa yi-AI njenge-WormGPT kanye ne-FraudGPT kukhombisa ukuthi "akuwona umsebenzi onzima ukwenza kabusha ubuchwepheshe obufanayo ngaphandle kwalezo zivikelo," kubhala u-Krishnan.

Eqinisweni, i-FraudGPT ne-WormGPT zisewubufakazi obengeziwe balokho omunye uchwepheshe wezokuphepha akubiza ngokuthi “i-generative AI jailbreaking for dummies,” lapho abalingisi ababi besebenzisa kabi izinhlelo zokusebenza ze-AI ukuze badlule. izimiso zokuziphatha ku-AI yokukhiqiza i-OpenAI ebikade ilwa nayo - impi ebilokhu ikhuphuka kakhulu.

"Kube umzabalazo oqhubekayo," kusho uPyry Avist, umsunguli kanye ne-CTO eHoxhunt. "Imithetho iyakhiwa, imithetho iyaphulwa, imithetho emisha iyakhiwa, leyo mithetho iyephulwa, njalo njalo."

Nakuba umuntu engeke “avele atshele i-ChatGPT ukuthi yenze i-imeyili egculisayo yobugebengu bokweba imininingwane ebucayi kanye nesifanekiso sokuvuna esithunyelwe sivela ku-CEO wakho,” othile “angenza sengathi uyi-CEO futhi abhale kalula i-imeyili ephuthumayo ethimbeni lezezimali alifune ukuthi lishintshe ukukhokhwa kwe-invoyisi. ," uthena.

Ukuvikela Ezinsongweni Ze-Cyber ​​ezinikwe amandla yi-AI

Ngempela, kulo lonke ibhodi, amathuluzi e-AI akhiqizayo anikeza izigebengu imisebenzi efanayo eyisisekelo ezinikeza ochwepheshe bezobuchwepheshe: "ikhono lokusebenza ngesivinini esikhulu nangesilinganiso," kusho u-Bambenek. "Abahlaseli manje sebengakwazi ukukhiqiza imikhankaso yobugebengu bokweba imininingwane ebucayi ngokushesha futhi baqale kanyekanye."

Njengoba ubugebengu bokweba imininingwane ebucayi buhlala bungenye yezindlela eziyinhloko lapho abahlaseli be-inthanethi baqala khona ukungena ohlelweni lwebhizinisi ukuze benze omunye umsebenzi omubi, kubalulekile ukusebenzisa ukuvikela okuvamile kukho. Lezi zivikelo zisengakwazi ukubona ubugebengu bokweba imininingwane ebucayi obunikwe amandla i-AI, futhi, okubaluleke nakakhulu, izenzo ezilandelwa ngumdlali osongelayo.

"Eqinisweni, lokhu akuguquli ukuguquguquka kokuthi umkhankaso wobugebengu bokweba imininingwane ebucayi uyini, noma umongo osebenza kuwo," kusho u-Bambenek. “Inqobo nje uma ungabhekani nobugebengu bokweba imininingwane ebucayi kusuka ku-akhawunti eyonakalisiwe, amasistimu ahlonishwayo asengakwazi ukubona ubugebengu bokweba imininingwane ebucayi kubathumeli bangempela, okungukuthi, izizinda ezibhalwe nge-typosquatted, ama-invoyisi avela kuma-akhawunti we-imeyili wewebhu wamahhala, njll.

Ukusebenzisa isu lokuvikela elijulile ngawo wonke ama-telemetry okuvikela atholakalayo ukuze kuhlaziywe ngokushesha futhi kungasiza izinhlangano zibone ukuhlasela kobugebengu bokweba imininingwane ebucayi ngaphambi kokuba abahlaseli bafake engozini isisulu futhi badlulele esigabeni esilandelayo sokuhlasela, esho.

"Abavikeli abadingi ukubona yonke into eyenziwa umhlaseli ochungechungeni olusongelayo, kufanele nje bathole okuthile ngaphambi kwezigaba zokugcina zokuhlasela - okungukuthi, i-ransomware noma i-data exfiltration. - ngakho ukuba nohlelo oluqinile lokuhlaziya idatha yezokuphepha kubalulekile,” kusho uBambenek.

Abanye ochwepheshe bezokuphepha nabo bakhuthaza ukusetshenziswa Amathuluzi okuphepha asekelwe ku-AI, izinombolo zazo zikhula, ukulwa ne-AI ephikisanayo, empeleni ukulwa nomlilo ngomlilo ukuze kuliwe nobunkimbinkimbi obukhulayo bendawo esongelayo.

• I-SEO Powered Content & PR Distribution. Khuliswa Namuhla.
• I-PlatoData.Network Vertical Generative Ai. Zinike Amandla. Finyelela Lapha.
• I-PlatoAiStream. I-Web3 Intelligence. Ulwazi Lukhulisiwe. Finyelela Lapha.
• I-PlatoESG. Izimoto / EVs, Ikhabhoni, I-CleanTech, Amandla, Environment, Ilanga, Ukuphathwa Kwemfucuza. Finyelela Lapha.
• I-BlockOffsets. Ukwenza kube yisimanje Ubunikazi be-Environmental Offset. Finyelela Lapha.
• Source: https://www.darkreading.com/threat-intelligence/fraudgpt-malicious-chatbot-for-sale-dark-web