Co esponding au ho : Eliel Kundai Zhuwankinyu
Copy igh © 2025 Au ho (s) e ain he copy igh o his a icle. This a icle is published unde he e ms o he C ea i e Commons A ibu ion License 4.0.
G aph-based secu i y models o AI-d i en da a s o age: A no el app oach o
p o ec ing classi ied documen s
Eliel Kundai Zhuwankinyu 1, *, Munashe Naph ali Mupa 2 and Syl es e Ta i enyika 2
1 Illinois Ins i u e o Technology.
2 Hul In e na ional Business School.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
Publica ion his o y: Recei ed on 25 Ma ch 2025; e ised on 05 May 2025; accep ed on 08 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.2.1631
Abs ac
Co po a e da a s o age sys ems a e suscep ible o cybe h ea s; hus, secu ing hem is a cen al p oblem in A i icial
In elligence (AI). G aph-Based Secu i y Models (GBSM) o m a eliable and scalable app oach o ein o cing
cybe secu i y. These models help o map ou ex ended cybe h ea s comp ehensi ely and acili a e enhanced h ea
iden i ica ion, anomaly de ec ion, and c yp og aphic in eg i y. Special emphasis has o do wi h in eg a ing AI wi h GBSM
as i enhances eal- ime anomaly de ec ion, au oma ed h ea esponse, c yp og aphic compu ing, and o he
app oaches ha make i a help ul solu ion o he secu ed handling o classi ied documen s in luid echnological
con ex s.
This wo k examines he speci ic p oblem o how adi ional app oaches o implemen ing in o ma ion secu i y a e
ine ec i e agains , o example, ze o-day exploi s and ad anced pe sis en h ea s. GBSM, he e o e, p o ides mo e
e sa ile secu i y measu es o de ence, which a e b ough abou by he cons an analysis o ela ionships be ween
di e en en i ies in di e en h ea ec o s. Addi ionally, ad anced elemen s o c yp og aphy key managemen and
decen alized blockchain amewo ks add mo e s eng h o he p o ec ion o iden i y and aluables, gi ing he
ad an age o a nea ly unal e able and anspa en access con ol, which a e emedies o mode n secu i y needs.
The p oposed s udy ocuses on in eg a ing GBSM and AI o de end dis ibu ed sys ems and cloud en i onmen s. I
explains how hese models allow o ganiza ions o map ou and ecognize h ea s and add ess hem be o e hey occu
in a decen alized en i onmen . Besides, he applica ion o g aph-based me hods in quan um-sa e c yp og aphy and
blockchain applica ions makes i possible o de elop p o ec ion agains no el h ea s in quan um compu ing and
ad e sa ial ac ions.
By using p og ams ha u ilize a i icial in elligence, his a icle explo es a p og essi e ou look on he issue o
cybe secu i y. He e, he sa es a place o he comp ehensi eness o u u e secu i y amewo ks en iched by AI, quan um
c yp og aphy, and GBSM, which should be sui able o u u e inc eased h ea s. Fu he mo e, he s udy ecommends
ha u u e wo ks o sol e he ou lined issues mus de elop adap i e AI models ha include pos -quan um
c yp og aphic me hods o p o ec ing da a when aced wi h new echnological h ea s.
Keywo ds: A i icial In elligence; Classi ied; Da a; G aph-based; Models
1. In oduc ion
G aph-Based Secu i y Models (GBSM) is an ad anced cybe secu i y pa adigm ha le e ages g aph- heo e ic s uc u es
o analyze, de ec , and mi iga e cybe h ea s in AI-d i en da a s o age en i onmen s. GBSM p o ides a scalable and
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1109
adap able secu i y amewo k ha enhances anomaly de ec ion, c yp og aphic in eg i y, and access con ol
mechanisms by ep esen ing cybe secu i y e en s as in e connec ed nodes wi hin a g aph s uc u e (Liu e al., 2022).
This app oach acili a es he iden i ica ion o complex a ack ec o s ha adi ional models s uggle o add ess.
Gi en he inc easing olume and complexi y o cybe h ea s, AI-d i en da a s o age secu i y is c i ical o p o ec ing
classi ied documen s. The in eg a ion o AI enhances secu i y ope a ions by enabling eal- ime de ec ion o anomalies,
au oma ing secu i y policy managemen , and imp o ing inciden esponse imes (Adenekan, 2024). When combined
wi h GBSM, AI models p o ide enhanced secu i y agains ad e sa ial a acks by le e aging g aph-based anomaly
de ec ion and ad anced c yp og aphic echniques (Al Siam e al., 2025).
T adi ional secu i y models ely on ule-based and signa u e-based de ec ion mechanisms, o en ailing agains
sophis ica ed h ea s such as ze o-day exploi s and ad anced pe sis en h ea s (Nagpu e, 2024). In con as , GBSM can
dynamically adap o e ol ing h ea landscapes by con inuously analyzing ela ionships be ween en i ies, making hem
supe io in eal- ime secu i y applica ions (Casas e al., 2023). As dis ibu ed ne wo ks and cloud-based in as uc u es
become mo e p e alen , he need o adap i e secu i y models is pa amoun . G aph-based app oaches enable
o ganiza ions o isualize a ack su aces mo e e ec i ely, allowing o p oac i e mi iga ion o h ea s in decen alized
a chi ec u es (Ejeo obi i e al., 2024). Addi ionally, GBSM con ibu es o c yp og aphic applica ions by enhancing key
managemen sys ems and de ec ing anomalies in enc yp ed a ic (Ta a da , 2024).
This pape explo es how GBSM imp o es secu i y amewo ks h ough AI-enhanced h ea de ec ion, c yp og aphic
applica ions, and eal- ime adap i e secu i y models. By in eg a ing AI wi h g aph-based s uc u es, GBSM o e s a
obus , scalable, and p oac i e app oach o secu ing classi ied documen s in an inc easingly complex digi al landscape
(Ejjami, 2024).
1.1. The Implica ions o G aph-Based Secu i y Models on T adi ional, Blockchain, and AI-Based
1.1.1. Key Secu i y
G aph-based app oaches in c yp og aphic models imp o e key secu i y by s uc u ing key managemen h ough
complex ela ionships and secu e mappings. Fuzzy g aph heo y has been explo ed o enhance key managemen
e iciency, enabling mo e secu e c yp og aphic sys ems (Singh, Khalid, and Nishad, 2024). A knowledge g aph-based
app oach also s eng hens secu i y policies by mapping access con ol me hods o secu e enc yp ed communica ion
(Chen e al., 2024). Addi ionally, blockchain-based c yp og aphic models in eg a e g aph secu i y o p e en
unau ho ized dec yp ion (Tsoulias e al., 2020). T ee-based c yp og aphic access con ol enhances dis ibu ed key
managemen , ensu ing secu i y in mul i-use en i onmen s (Alde man, Fa ley, and C amp on, 2017). Such app oaches
p o ide scalable, a ack- esis an c yp og aphic models.
Mo eo e , G aph-Based Secu i y Models (GBSMs) enhance blockchain-based key secu i y by p o iding s uc u ed
c yp og aphic me hods o decen alized au hen ica ion and access con ol (Wan e al., 2024). The decen alized and
immu able na u e o blockchain aligns wi h g aph-based secu i y, ensu ing anspa en and ampe -p oo key
managemen (De Alwis, Pham, and Liyanage, 2022). In Indus y 4.0, blockchain-in eg a ed GBSMs secu e ansac ions
and en o ce c yp og aphic policies (Bha acha ya e al., 2021). These app oaches op imize key secu i y while
suppo ing scalable, AI-d i en h ea de ec ion in eme ging echnologies (Po ambage and Liyanage, 2023).
AI-enhanced c yp og aphic key managemen e olu ionizes access con ol by imp o ing secu i y, e iciency, and
scalabili y in AI-d i en s o age en i onmen s. Adap i e AI-d i en enc yp ion dynamically adjus s key managemen
s a egies, educing ulne abili ies inhe en in s a ic c yp og aphic models (Ahmad e al., 2025). AI-d i en iden i y and
access managemen (IAM) u he s eng hens au hen ica ion p o ocols, minimizing unau ho ized access isks
(Rehman and Ali, 2024). AI-in eg a ed blockchain solu ions p o ide addi ional secu i y laye s, ensu ing decen alized
and immu able key s o age (Ruzbahani, 2024).
G aph secu i y models s eng hen ze o- us a chi ec u es by mi iga ing key comp omise isks h ough con inuous
access alida ion and anomaly de ec ion (Ahmadi, 2024). These models implemen mic o-segmen a ion, p e en ing
la e al mo emen o h ea s in case o c eden ial b eaches (Ghasemshi azi, Shi ani, and Alipou , 2023). G aph-based
analy ics dynamically moni o au hen ica ion beha io s, imp o ing key secu i y h ough au oma ed h ea esponses
(Syed e al., 2022). Addi ionally, hese models p o ide obus c yp og aphic key dis ibu ion, p e en ing inside h ea s
by alida ing en i ies based on eal- ime isk assessmen s (Ale izos, Eiza, and Ta, 2022).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1110
G aph-based key secu i y also o e s enhanced scalabili y and esilience compa ed o adi ional Public Key
In as uc u e (PKI) by u ilizing decen alized us models (Gu u e al., 2023). A case s udy in inancial se ices
demons a ed ha g aph-based secu i y educed c yp og aphic o e head while main aining high au hen ica ion
in eg i y (Kahmann e al., 2023). Unlike PKI, which elies on cen alized Ce i ica e Au ho i ies (CAs), g aph-based
me hods o e mo e obus esis ance o quan um-based h ea s (Salama, Shams, and Bha naga , 2023). These models
enhance key dis ibu ion e iciency and secu i y lexibili y in decen alized ne wo ks (Maldonado-Ruiz, To es, and El
Madhoun, 2022).
1.2. Implica ions o G aph-Based Secu i y on Quan um C yp og aphy and AI in Th ea Analysis
Quan um c yp og aphy (QC) le e ages quan um mechanics o enhance secu e communica ion, pa icula ly h ough
Quan um Key Dis ibu ion (QKD), which ensu es uncondi ional secu i y agains compu a ional a acks (Sood, 2024).
Howe e , quan um compu e s pose a signi ican h ea o classical c yp og aphic sys ems, as Sho ’s algo i hm enables
e icien ac o iza ion o la ge numbe s, b eaking RSA and ECC enc yp ion (Hosseini and Pila am, 2024). Pos -quan um
c yp og aphy (PQC) aims o de elop quan um- esis an algo i hms, wi h la ice-based and hash-based c yp og aphy
eme ging as p omising al e na i es (Li e al., 2023). The ansi ion o PQC p esen s challenges, including
s anda diza ion, pe o mance ade-o s, and in as uc u e adap a ion (Sha ma e al., 2023).
G aph-based secu i y models p o ide enhanced esilience agains quan um a acks by s uc u ing key managemen and
au hen ica ion mechanisms wi h quan um- esis an c yp og aphic p o ocols (Oli a delMo al and deMa i iOlius, 2024).
These models in eg a e wi h pos -quan um c yp og aphy by employing hash-based and la ice-based enc yp ion
schemes o p e en quan um-based key comp omise (Singamaneni and Muhammad, 2024). Addi ionally, g aph
s uc u es imp o e dis ibu ed ledge secu i y, ensu ing c yp og aphic ope a ions emain secu e in quan um
en i onmen s (Xu e al., 2023). Thei ole in secu ing key exchanges and ein o cing c yp og aphic us amewo ks
makes hem c ucial o u u e quan um-sa e in as uc u e (Thanalakshmi e al., 2021).
AI-d i en secu i y enhances dis ibu ed ne wo k p o ec ion by employing eal- ime anomaly de ec ion and au oma ed
h ea esponse mechanisms (Ka i ha and Thejas, 2024). Deep lea ning models analyze ne wo k beha io o iden i y
complex h ea ec o s, enabling p oac i e secu i y measu es (Tan e al., 2024). AI-d i en mapping o cybe h ea s
allows o he iden i ica ion o a ack s a egies and imp o es adap i e secu i y policies (Pa acha e al., 2024). These
solu ions educe esponse imes and mi iga e la ge-scale dis ibu ed denial-o -se ice (DDoS) a acks in decen alized
sys ems (Zacha is, Ka os, and Pa sakis, 2024).
G aph- heo y-based anomaly de ec ion enhances AI-powe ed Secu i y Ope a ions Cen e s (SOCs) by mapping
cybe secu i y h ea s h ough g aph analy ics, educing dwell ime in a ack de ec ion (Rahman, 2024). These
echniques u ilize g aph s uc u es o ep esen complex secu i y e en s, imp o ing AI-d i en h ea co ela ion and
p edic i e analy ics (El Azzaoui e al., 2020). G aph-based anomaly de ec ion enhances au oma ed secu i y esponses,
enabling SOCs o p io i ize and mi iga e cybe inciden s e icien ly (Md Sha ia Sozol e al., 2024). By in eg a ing AI wi h
g aph algo i hms, SOCs imp o e si ua ional awa eness, p oac i ely de ec ing and p e en ing ad anced pe sis en
h ea s in eal- ime.
G aph-based h ea isualiza ion enhances Cybe Th ea In elligence (CTI) by enabling secu i y analys s o map a ack
pa e ns and ela ionships among cybe h ea s (Jia e al., 2025). AI-d i en CTI pla o ms u ilize g aph analy ics o
ex ac insigh s om s uc u ed and uns uc u ed da a, imp o ing si ua ional awa eness (B a sas, Anas asiadis, and
Angelidis, 2024). Ad anced pe sis en h ea (APT) de ec ion is signi ican ly imp o ed h ough g aph-based algo i hms
ha analyze h ea in elligence epo s (Gulbay and Demi ci, 2024). These me hods au oma e a ack ec o co ela ion,
educing de ec ion ime and enabling p oac i e cybe secu i y s a egies (Li e al., 2023).
AI-powe ed eal- ime h ea de ec ion in dis ibu ed ne wo ks also enhances cybe secu i y by iden i ying anomalies
and mi iga ing a acks be o e hey escala e (Rehman and Weng, 2025). Fede a ed lea ning models imp o e dis ibu ed
h ea de ec ion by aining AI algo i hms wi hou exposing sensi i e da a (Anandha aj, 2024). AI-d i en pla o ms
enable con inuous ne wo k moni o ing and apid esponse o h ea s like Dis ibu ed Denial-o -Se ice (DDoS) a acks
(Mi za and Huide , 2024). These solu ions signi ican ly enhance cybe secu i y esilience by educing esponse imes
and au oma ing secu i y p o ocols (Abdel-Wahid, 2024).
1.3. Implica ions o Big Da a, IoT Secu i y, and Fu u e Di ec ions
G aph-based secu i y models p o ide scalable and e icien mechanisms o secu ing la ge-scale big da a in as uc u es
by enabling eal- ime anomaly de ec ion and a ack co ela ion (Win, Tian ield, and Mai , 2017). These models use
g aph-based e en co ela ion o analyze complex a ack pa e ns, enhancing cybe secu i y in i ualized cloud
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1111
in as uc u es. Addi ionally, g aph heo y suppo s big da a analy ics in IoT en i onmen s by acili a ing adap i e
secu i y policies o he e ogeneous de ices (Ra ho e e al., 2021).
The inc easing complexi y o IoT ne wo ks equi es obus secu i y amewo ks o mi iga e cybe h ea s. G aph-based
echniques e icien ly iden i y IoT ulne abili ies by analyzing abno mal a ic pa e ns and de ec ing ou lie s in
ne wo k beha io (Gao e al., 2023). Addi ionally, hese models enhance su i abili y assessmen s by op imizing
in usion p e en ion s a egies o IoT applica ions (Shakho and Koo, 2021). A no el g aph-based app oach has also
been applied o IoT bo ne de ec ion, imp o ing esilience agains dis ibu ed denial-o -se ice (DDoS) a acks (Nguyen,
Ngo, and Le, 2020). These secu i y ad ancemen s demons a e he c i ical ole o g aph-based secu i y models in
p o ec ing la ge-scale in as uc u es.
Mo eo e , AI-d i en IoT da a managemen p esen s signi ican p i acy conce ns, pa icula ly in secu ing sensi i e da a
om unau ho ized access and misuse (Ma engo, 2024). The in eg a ion o AI in IoT sys ems enables eal- ime da a
p ocessing bu inc eases exposu e o cybe h ea s and da a b eaches. T anspa en da a go e nance amewo ks a e
necessa y o ensu e compliance wi h global p i acy egula ions such as GDPR and CCPA (Ma engo, 2024). P i acy-
p ese ing AI echniques, including di e en ial p i acy and homomo phic enc yp ion, a e being adop ed o mi iga e
hese isks while main aining e icien IoT da a ope a ions (Cas o, Deng, and Pa k, 2023).
G aph AI-d i en anomaly de ec ion enhances IoT secu i y by iden i ying cybe h ea s and ne wo k anomalies h ough
ad anced pa e n ecogni ion echniques (Salem, Said, and Nou , 2024). These models le e age G aph Neu al Ne wo ks
(GNNs) o de ec eal- ime secu i y h ea s and au oma e in usion de ec ion (Ejeo obi i, Vic o -Igun, and Okoye, 2024).
AI-enhanced anomaly de ec ion amewo ks signi ican ly imp o e IoT eliabili y, educing alse posi i es in cybe
h ea de ec ion (Wajid and Sans, 2024). These solu ions o e scalable, p oac i e secu i y measu es essen ial o he
g owing IoT ecosys em.
Au onomous secu i y amewo ks o IoT a e eme ging as a c i ical esea ch di ec ion o add ess he inc easing
complexi y o cybe h ea s. AI-d i en adap i e secu i y models a e being de eloped o enhance h ea in elligence and
au oma ed esponse mechanisms in 5G-enabled IoT ecosys ems (Abie and Pi bhulal, 2024). Decen alized secu i y
app oaches, such as blockchain-in eg a ed AI amewo ks, a e imp o ing IoT de ice au hen ica ion and da a in eg i y
(Figuei edo e al., 2022). Fu u e ad ancemen s will also ocus on au onomous in usion de ec ion using machine
lea ning o mi iga e eal- ime a acks (Akhunzada, Al-Shamayleh, and Zeadally, 2024). These amewo ks p omise
scalable, sel -sus aining cybe secu i y o nex -gene a ion IoT ne wo ks.
1.4. Rela ionship wi h IoMT (In e ne o Medical Things) Secu e Da a Managemen F amewo k
G aph-based secu i y models play a c ucial ole in p o ec ing In e ne o Medical Things (IoMT) de ices by p o iding
scalable and adap i e secu i y amewo ks. These models use g aph analy ics o de ec and p e en cybe h ea s by
mapping a ack ec o s and iden i ying ulne abili ies in eal- ime (Lo ù, 2022). By le e aging g aph-based anomaly
de ec ion, IoMT ne wo ks can p oac i ely mi iga e isks associa ed wi h unau ho ized access and da a b eaches
(Ka aa slan and Konacaklı, 2021). AI-enhanced g aph secu i y u he imp o es eal- ime h ea in elligence,
au oma ing de ec ion mechanisms o sa egua d IoMT de ices om eme ging cybe h ea s (Wen, Shukla, and Ka ,
2025).
AI-powe ed secu i y g aphs enhance heal hca e da a p i acy by ensu ing obus enc yp ion, secu e access con ol, and
compliance wi h egula o y s anda ds. G aph neu al ne wo ks (GNNs) suppo he implemen a ion o decen alized
p i acy amewo ks, educing he isk o cen alized da a b eaches (Singh and Siddiqui, 2024). These models allow o
e icien anonymiza ion o pa ien eco ds, enabling p i acy-p ese ing AI applica ions in elec onic heal h eco ds
(EHRs) (Khalid e al., 2023). Fu he mo e, AI-d i en p i acy-p ese ing echniques, such as ede a ed lea ning, ensu e
secu e medical da a p ocessing wi hou comp omising pa ien con iden iali y (Majeed, Khan, and Hwang, 2022). These
ad ancemen s highligh he ans o ma i e impac o AI-powe ed secu i y g aphs in sa egua ding sensi i e heal hca e
in o ma ion.
G aph-based h ea in elligence enhances a ack ec o analysis in In e ne o Medical Things (IoMT) en i onmen s by
mapping cybe h ea s and iden i ying ulne abili ies in heal hca e ne wo ks (Naghib, Gha ehchopogh, and Zamani a ,
2025). These models de ec and isualize a ack pa hs in IoMT sys ems, mi iga ing isks posed by weak au hen ica ion,
unenc yp ed da a ansmission, and ou da ed secu i y p o ocols (Ghodsizad, 2024). G aph-based secu i y models
e ec i ely coun e man-in- he-middle (MITM) and Sybil a acks in IoMT by le e aging machine lea ning o adap i e
anomaly de ec ion (Nagamani and Kuma , 2024). Fu he mo e, g aphical secu i y modeling (GSM) has been
implemen ed o assess and p e en a ack p opaga ion ac oss in e connec ed IoMT de ices (AboulEla e al., 2024).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1112
A case s udy on hospi al ne wo k secu i y highligh s he e ec i eness o g aph-based secu i y models in p e en ing
cybe h ea s in heal hca e cloud s o age (Ra i, Pham, and Alazab, 2022). AI-d i en secu i y g aphs s eng hen
heal hca e da a ansmission secu i y by p oac i ely iden i ying suspicious ac i i y and op imizing access con ol
policies (Naph ali Mupa e al., 2025). Cloud-based IoMT amewo ks in eg a e g aph analy ics o p o ec pa ien eco ds
and enc yp ed medical da a om ansomwa e and unau ho ized access (P asad e al., 2022). This esea ch unde sco es
he necessi y o g aph-based secu i y in mode n heal hca e in as uc u es.
G aph-based secu i y amewo ks also enhance egula o y compliance by au oma ing da a p o ec ion and ensu ing
adhe ence o p i acy laws such as HIPAA and GDPR (Ka alka and Medi skos, 2024). These models use g aph-based isk
managemen o de ec ulne abili ies in da a p ocessing and en o ce eal- ime secu i y policies (Alja ah, Che bal, and
Mashaleh, 2024). G aph p ope y analysis has been applied o p i acy h ea modeling, imp o ing compliance
au oma ion in cloud-based heal hca e applica ions (Kunz, Weiss, and Schneide , 2023). Addi ionally, knowledge g aphs
enable s uc u ed da a access con ol, enhancing audi abili y and educing compliance iola ions (Sangee ha,
Sel a a hi, and Ma hi anan, 2024). These ad ancemen s ensu e ha o ganiza ions e icien ly main ain egula o y
secu i y and p i acy s anda ds.
1.5. The Impac o AI-D i en Enhancemen s in Cloud Compu ing Secu i y
AI-powe ed g aph secu i y enhances cloud compu ing en i onmen s by enabling eal- ime anomaly de ec ion and
adap i e h ea esponse mechanisms (Moo hy and Jaganna h, 2024). G aph-based models imp o e cybe secu i y in
cloud ne wo ks by analyzing connec ions be ween digi al asse s, helping de ec and neu alize cybe h ea s e icien ly
(Ullah, Kamal, and Asi , 2024). Addi ionally, AI-d i en secu i y g aphs suppo au oma ed compliance moni o ing,
educing human in e en ion while ensu ing egula o y adhe ence (Ankalaki e al., 2025).
The syne gy be ween ML, Gene a i e AI, and g aph-based cybe secu i y is ans o ming h ea in elligence. Machine
lea ning models use g aph-based analy ics o unco e complex a ack pa e ns, s eng hening in usion de ec ion
sys ems (IDS) (Zhang e al., 2024). Gene a i e AI u he enhances cybe secu i y by p edic ing po en ial a ack ec o s
and op imizing secu i y esponses (Al Siam e al., 2025). These AI-d i en echnologies, when in eg a ed wi h g aph-
based secu i y amewo ks, enable scalable and sel -lea ning secu i y in as uc u es capable o p oac i e cybe
de ense (Sindi amu y and P abaga an, 2025).
Ze o- us a chi ec u e (ZTA) is e olu ionizing AI-d i en cloud secu i y by elimina ing implici us and en o cing
s ic access con ols based on con inuous e i ica ion (Ahmadi, 2025). Unlike pe ime e -based models, ZTA in eg a es
AI-d i en au hen ica ion and g aph-based policy en o cemen o mi iga e unau ho ized access isks (Xun e al., 2025).
AI-d i en ZTA enhances eal- ime h ea de ec ion by le e aging g aph analy ics o analyze a ack pa e ns and p edic
in usion a emp s (Je n e al., 2025). Mo eo e , in eg a ing adap i e mul i- ac o au hen ica ion (MFA) wi hin ZTA
models ensu es imp o ed secu i y esilience agains sophis ica ed cybe h ea s (Nagpu e, 2024).
Supply chain a acks in mul i-cloud en i onmen s a e becoming mo e p e alen , necessi a ing AI-d i en g aph-based
secu i y app oaches (Joshi, 2024). G aph-based h ea in elligence maps ulne abili ies ac oss in e connec ed cloud
p o ide s, iden i ying po en ial b each poin s be o e exploi a ion (Hassan, Nizam-Uddin, and Quddus, 2024). AI-d i en
opology g aph-based anomaly de ec ion (TOGBAD) models u he enhance supply chain secu i y by iden i ying
anomalies in da a lows and ansac ion logs (Ge, 2024). This in eg a ion o AI, ZTA, and g aph-based secu i y ensu es
a esilien de ense agains e ol ing cybe h ea s in mul i-cloud in as uc u es.
AI-enhanced g aph-based secu i y is ans o ming cloud secu i y in AWS, Azu e, and Google Cloud by imp o ing isk
mi iga ion and eal- ime anomaly de ec ion (Xun e al., 2025). AWS u ilizes opology g aph-based anomaly de ec ion
(TOGBAD) o iden i y secu i y h ea s wi hin i s in as uc u e, s eng hening access con ol mechanisms (Hassan,
Nizam-Uddin, and Quddus, 2024). Simila ly, Azu e in eg a es AI-d i en schedule s and g aph-based lineage in e ence
models o enhance secu i y isibili y and ensu e cloud wo kload p o ec ion (Naph ali Mupa e al., 2025). Google Cloud
le e ages dis ibu ed acing and g aph-based de ec ion me hods o moni o cloud se ice in e ac ions and au oma e
secu i y esponses (Rallabandi, 2024). These ad ancemen s in AI-powe ed cloud secu i y unde sco e he g owing
eliance on g aph-based h ea in elligence.
2. AI-Powe ed Th ea Hun ing in SAP and ERP En i onmen s
AI-d i en cybe secu i y enhances en e p ise en i onmen s by imp o ing h ea de ec ion, aud p e en ion, and
secu i y au oma ion (Moo e and Rou hu, 2024). En e p ises a e le e aging AI-powe ed secu i y solu ions o enhance
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1113
isk assessmen , in eg a ing machine lea ning algo i hms wi h cybe secu i y p o ocols o iden i y po en ial secu i y
b eaches in eal ime (Lun o skyy e al., 2024).
G aph-based anomaly de ec ion plays a i al ole in secu ing SAP and ERP sys ems by mapping ne wo k beha io ,
iden i ying abno mal pa e ns, and mi iga ing da a b eaches (S u ano, 2024). By in eg a ing g aph-based machine
lea ning, ERP secu i y amewo ks can de ec audulen ac i i ies wi hin en e p ise esou ce planning en i onmen s
(Eliel e al., 2025). This ensu es o ganiza ions main ain obus cybe secu i y pos u es and comply wi h egula o y
equi emen s.
Inside h ea s and Ad anced Pe sis en Th ea s (APTs) p esen majo secu i y isks in En e p ise Resou ce Planning
(ERP) amewo ks, o en bypassing adi ional secu i y con ols o exploi p i ileged access (Rod igues, 2019). AI-
d i en secu i y models use beha io al analy ics o de ec anomalous use ac i i ies, p e en ing unau ho ized access o
c i ical ERP da a. G aph-based anomaly de ec ion plays a c ucial ole in iden i ying audulen ansac ions,
sa egua ding inancial and ope a ional da a om manipula ion (Tan e al., 2025).
A case s udy on SAP secu i y e eals ha g aph-based h ea in elligence e ec i ely de ec s p i ilege escala ion by
mapping access con ol ela ionships and moni o ing de ia ions om no mal usage pa e ns (Rod igues, 2019). By
analyzing ole inhe i ance and access hie a chies, o ganiza ions can p oac i ely mi iga e unau ho ized p i ilege
escala ion a emp s in ERP sys ems. This g aph-based secu i y app oach enhances he isibili y o a ack pa hs, allowing
o apid esponse o eme ging h ea s (Mehmood e al., 2023).
2.1. Secu i y Th ea s in AI-D i en Cloud En i onmen s
AI-d i en cloud en i onmen s ace a g owing numbe o secu i y h ea s, including ad e sa ial a acks, da a poisoning,
and model in e sion. Ad e sa ial a acks manipula e inpu da a o decei e AI models, leading o inco ec p edic ions
and po en ial sys em comp omise (Zhuwankinyu e al., 2023). Da a poisoning in oduces co up ed aining da a o
al e AI model beha io , educing accu acy and inc easing ulne abili ies (Reddy, Konkimalla, & Raja am, 2022). Model
in e sion a acks ex ac sensi i e da a om ained models, aising p i acy conce ns (Alaca, Celık, & Goel, 2023).
Con aine ized AI en i onmen s, pa icula ly hose using Kube ne es, p esen secu i y isks such as con aine escapes,
whe e malicious ac o s gain access o he hos sys em, and Kube ne es clus e a acks, a ge ing miscon igu a ions and
p i ilege escala ion (Mi opoulou, Kokkinos, & Soumplis, 2024). Cloud-based AI secu i y is u he h ea ened by
unau ho ized access, API abuse, and da a leakage, pa icula ly in mul i-cloud deploymen s (Wijenayake & Henna, 2023).
G aph-Based Secu i y Models (GBSM) p o ide a s uc u al app oach o iden i ying ulne abili ies in AI-d i en
wo k lows by mapping a ack su aces and acking malicious ac i i ies (G a a, Deshpande, & Lopes, 2024). G aph
analy ics enhances p edic i e h ea de ec ion, imp o ing AI model esilience in mul i-cloud and con aine ized AI
deploymen s (Nagpu e, 2024).
GBSM acili a es a ack pa h isualiza ion in complex AI wo k lows, allowing o p oac i e secu i y measu es in cloud
in as uc u es (Khan, Ma skin, & P odan, 2024). By in eg a ing knowledge g aphs and machine lea ning, secu i y
models de ec anomalies in AI model in e ac ions and p e en po en ial ad e sa ial h ea s (Zhong e al., 2024). These
solu ions signi ican ly enhance AI secu i y by moni o ing con aine ized applica ions, en o cing ze o- us
au hen ica ion, and p e en ing model manipula ion a acks (Nguyen, Zhu, & Liu, 2022).
2.2. Inco po a ing Ad anced Secu i y Techniques in G aph-Based Models
Di e en ial p i acy (DP) is a echnique ha ensu es AI models do no expose indi idual da a poin s, main aining s ong
p i acy gua an ees (Luo e al., 2024). I is widely used in cloud-based AI pla o ms, such as Google AI, Mic oso Azu e
ML, and AWS Sagemake , o p e en e-iden i ica ion a acks du ing model aining (Qiu e al., 2022). G aph-Based
Secu i y Models (GBSM) ack da a anonymiza ion by applying di e en ial p i acy echniques o enc yp ed g aph
nodes, ensu ing obus p i acy-p ese ing AI ope a ions (Fu e al., 2023).
Fede a ed lea ning (FL) enables decen alized model aining ac oss mul iple de ices wi hou sha ing aw da a,
mi iga ing p i acy isks (Mansou Baha , Fe ahi & Messai, 2024). Howe e , FL emains ulne able o secu i y h ea s
such as model in e sion and poisoning a acks, whe e ad e sa ies manipula e aining upda es o in e sensi i e da a
(Han e al., 2024). GBSM enhances FL secu i y by iden i ying anomalous pa e ns in model upda es, lagging po en ially
comp omised nodes in decen alized AI wo k lows (Luo e al., 2023). In p ac ice, FL combined wi h g aph-based
anomaly de ec ion is applied in cloud en i onmen s like GCP Ve ex AI and Azu e AI o secu e collabo a i e AI aining
(Pauu e al., 2023).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1114
Enc yp ion plays a c i ical ole in AI secu i y, wi h homomo phic enc yp ion allowing compu a ions on enc yp ed da a,
p ese ing p i acy (Zhang, 2023). End- o-end enc yp ion ensu es secu e communica ion in AI wo k lows by p e en ing
unau ho ized access (Ye e al., 2023). GBSM imp o es enc yp ion-based secu i y policies by mapping c yp og aphic
us s uc u es, enhancing compliance in cloud-based AI models (Fu e al., 2023). In Google Cloud Pla o m (GCP),
con iden ial compu ing u ilizes g aph-based secu i y o ein o ce AI model p o ec ion, ensu ing da a con iden iali y in
aining and deploymen (K Zhang, 2023).
Da a masking plays a c ucial ole in secu ing sensi i e AI aining da ase s by ob usca ing pe sonally iden i iable
in o ma ion (PII) while main aining analy ical in eg i y (Chen e al., 2024). G aph-Based Secu i y Models (GBSM)
enhance his by acking masked da a lows o p e en unau ho ized exposu e in AI wo k lows (Jia e al., 2024). In cloud
AI applica ions, da a masking is i al in heal hca e and inancial se ices o p o ec sensi i e eco ds while enabling
p edic i e analy ics (Nandan, Mi a & De, 2025).
Ze o T us A chi ec u e (ZTA) also ensu es AI secu i y h ough con inuous e i ica ion and leas -p i ilege access,
elimina ing implici us in cloud en i onmen s (Xin e al., 2025). G aph-based au hen ica ion s eng hens ZTA by
moni o ing access pa hways and p e en ing unau ho ized la e al mo emen (Gambo & Almulhem, 2025). Google’s
BeyondCo p implemen s ZTA p inciples o AI secu i y, in eg a ing g aph-based access con ol o p o ec AI-d i en
cloud se ices (Ye e al., 2024).
2.3. P ac ical Implemen a ion: Secu ing AI Con aine s and Wo k lows wi h G aph-Based Secu i y
AI con aine ized en i onmen s, such as Kube ne es, a e suscep ible o ulne abili ies including supply chain a acks
and p i ilege escala ion (A huko ale e al., 2025). G aph-Based Secu i y Models (GBSM) can mi iga e hese h ea s by
employing a ack g aph isualiza ion o de ec malicious ac i i y in AI clus e s (Mi a e al., 2024). In cloud-based
Kube ne es deploymen s, in eg a ing GBSM enhances secu i y h ough anomaly de ec ion and p edic i e h ea
moni o ing (Nguyen, Zhu & Liu, 2022).
Google Cloud’s An hos and AI Pla o m le e age g aph-based secu i y models o s eng hen AI pipeline secu i y (Pa el
e al., 2024). GBSM is pa icula ly use ul o g aph-enhanced in usion de ec ion, which sa egua ds AI wo k lows om
da a ex il a ion and ad e sa ial model manipula ions (Rallabandi, 2024). By in eg a ing g aph analy ics wi h Google
Cloud Func ions, hese models enhance AI execu ion while main aining con aine -based isola ion (Wijenayake & Henna,
2023).
In e ms o u u e ends, quan um-sa e c yp og aphy is an eme ging end in AI-d i en cloud secu i y, add essing he
isks posed by quan um compu ing o adi ional enc yp ion me hods (G a a, Deshpande & Lopes, 2024). Howe e ,
implemen ing g aph-based secu i y a scale is challenging due o he complexi y o handling la ge-scale g aph
compu a ions and eal- ime h ea de ec ion (Zhong e al., 2024). The need o AI-d i en secu i y au oma ion in g aph-
based h ea in elligence is g owing, as manual h ea esponse is ine icien in dynamic cloud en i onmen s (Ramya,
Sme a & Sandeep, 2025). AI-based anomaly de ec ion enhances p edic i e cybe secu i y by iden i ying isks be o e
exploi a ion occu s (Adenekan, 2024).
3. Recommenda ion and Conclusion
This a icle explo ed he signi icance o G aph-Based Secu i y Models (GBSMs) in AI-d i en da a s o age, highligh ing
hei ole in mi iga ing cybe h ea s, p e en ing key comp omises, and enhancing c yp og aphic applica ions (Paul,
2024). We examined AI-enhanced c yp og aphic key managemen , g aph-based anomaly de ec ion, and he in eg a ion
o quan um-sa e c yp og aphy o secu e sensi i e da a in en e p ise and cloud en i onmen s. The s udy u he
analyzed case s udies on AWS, Azu e, and Google Cloud, demons a ing how g aph secu i y enhances mul i-cloud
esilience, IoT, and ERP secu i y (Rachid Ejjami, 2024).
To add ess e ol ing cybe h ea s, o ganiza ions should in eg a e A i icial In elligence (AI), Quan um C yp og aphy,
and G aph-Based Secu i y in o cybe secu i y amewo ks (Dhanamma Jagli, 2024). AI-d i en g aph analy ics should be
le e aged o eal- ime a ack de ec ion, inside h ea mi iga ion, and supply chain secu i y (Kel in O abo e al., 2024).
Fu u e esea ch should ocus on adap i e AI models ha in eg a e pos -quan um c yp og aphic echniques, ensu ing
secu i y agains eme ging quan um compu ing h ea s.
Fu he ad ancemen s in g aph-based cybe secu i y amewo ks should emphasize au onomous secu i y decision-
making, educing manual in e en ion. Fede a ed lea ning models, ze o- us a chi ec u es, and blockchain-enhanced
access con ol should be in eg a ed wi h g aph-d i en isk assessmen o secu e decen alized in as uc u es (F eed
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1115
& Jackson, 2022). Addi ionally, eal- ime policy en o cemen using AI-based egula o y compliance moni o ing will
s eng hen secu i y go e nance (Joshi, 2025).
F om a policy pe spec i e, go e nmen s and indus y leade s mus es ablish s anda dized e hical AI secu i y guidelines
o ensu e esponsible implemen a ion. Regula o y bodies such as HIPAA, GDPR, and NIST should e ine amewo ks o
add ess AI-d i en cybe secu i y isks in c i ical sec o s (Lund e al., 2025). E hical AI secu i y measu es mus p io i ize
anspa ency, ai ness, and bias mi iga ion, ensu ing obus and us wo hy cybe secu i y ecosys ems. G aph-based
secu i y models will con inue o shape he u u e o AI-d i en cybe secu i y, o e ing scalable, in elligen , and adap i e
secu i y solu ions o complex cybe h ea s (NIST, 2021).
Compliance wi h e hical s anda ds
Disclosu e o con lic o in e es
No con lic o in e es o be disclosed.
Re e ences
[1] Abdel-Wahid, T. (2024) 'AI-Powe ed Cloud Secu i y: A S udy on he In eg a ion o A i icial In elligence and
Machine Lea ning o Imp o ed Th ea De ec ion and P e en ion', Resea chGa e. A ailable a :
h ps://www. esea chga e.ne /p o ile/Resea ch-Schola -Ii/publica ion/383095008_AI-
POWERED_CLOUD_SECURITY_A_STUDY_ON_THE_INTEGRATION_OF_ARTIFICIAL_INTELLIGENCE_AND_MACHI
NE_LEARNING_FOR_IMPROVED_THREAT_DETECTION_AND_PREVENTION/links/66bc8a e145 4d355357 81b
/AI-POWERED-CLOUD-SECURITY-A-STUDY-ON-THE-INTEGRATION-OF-ARTIFICIAL-INTELLIGENCE-AND-
MACHINE-LEARNING-FOR-IMPROVED-THREAT-DETECTION-AND-PREVENTION.pd (Accessed: 12 Ma ch
2025).
[2] Abie, H. and Pi bhulal, S. (2024) 'Au onomous Adap i e Secu i y F amewo k o 5G-Enabled IoT', a Xi . DOI:
10.48550/a Xi .2406.03186 (Accessed: 12 Ma ch 2025).
[3] AboulEla, S., Ib ahim, N., Shehmi , S., Yada , A., and Kashe , R. (2024) 'Na iga ing he cybe h ea landscape: An
in-dep h analysis o a ack de ec ion wi hin IoT ecosys ems', MDPI AI. DOI: 10.3390/ai5020037 (Accessed: 12
Ma ch 2025).
[4] Adenekan, T.K. (2024) 'Op imized AI and G aph-Regula ized Neu al Ne wo ks o Cybe Secu i y and AIOps in
IoT', Resea chGa e. A ailable a : h ps://www. esea chga e.ne /p o ile/Tobiloba-
Adenekan/publica ion/386441217_Op imized_AI_and_G aph-
Regula ized_Neu al_Ne wo ks_ o _Cybe _Secu i y_and_AIOps_in_IoT/links/675176d1e17e 87d1de64998/Op i
mized-AI-and-G aph-Regula ized-Neu al-Ne wo ks- o -Cybe -Secu i y-and-AIOps-in-IoT.pd [Accessed 14
Ma ch 2025].
[5] Adenekan, T.K. (2024) Op imized AI and G aph-Regula ized Neu al Ne wo ks o Cybe Secu i y and AIOps in
IoT. Resea chGa e. A ailable a : h ps://www. esea chga e.ne /p o ile/Tobiloba-
Adenekan/publica ion/386441217_Op imized_AI_and_G aph-
Regula ized_Neu al_Ne wo ks_ o _Cybe _Secu i y_and_AIOps_in_IoT/links/675176d1e17e 87d1de64998/Op i
mized-AI-and-G aph-Regula ized-Neu al-Ne wo ks- o -Cybe -Secu i y-and-AIOps-in-IoT.pd (Accessed: 12
Ma ch 2025).
[6] Ahmad, A., Rehman, A.U., Ghani, M.U., Nasim, F., and Naseem, S. (2025) ‘An In-Dep h Compa a i e Analysis o
T adi ional s AI-Enhanced Enc yp ion Algo i hms’, Al-Aasa Jou nal. A ailable a : h p://al-
aasa .com/index.php/Jou nal/a icle/ iew/147 (Accessed: 12 Ma ch 2025).
[7] Ahmadi, S. (2024) Ze o us a chi ec u e in cloud ne wo ks: Applica ion, challenges and u u e oppo uni ies.
HAL Science. A ailable a : h ps://hal.science/hal-04456272/ (Accessed: 12 Ma ch 2025).
[8] Ahmadi, S. (2025) 'Au onomous Iden i y-Based Th ea Segmen a ion in Ze o T us A chi ec u es', a Xi . DOI:
10.48550/a Xi .2501.06281 (Accessed: 12 Ma ch 2025).
[9] Akhunzada, A., Al-Shamayleh, A.S., and Zeadally, S. (2024) 'Design and pe o mance o an AI-enabled h ea
in elligence amewo k o IoT-enabled au onomous ehicles', ScienceDi ec . A ailable a :
h ps://www.sciencedi ec .com/science/a icle/pii/S0045790624005366 (Accessed: 12 Ma ch 2025).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1116
[10] Al Siam, A., Alazab, M., Awajan, A. and Fa uqui, N. (2025) ‘A Comp ehensi e Re iew o AI's Cu en Impac and
Fu u e P ospec s in Cybe secu i y’, IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10836696/ (Accessed: 12 Ma ch 2025).
[11] Alaca, Y., Celık, Y. and Goel, S. (2023) 'Anomaly de ec ion in cybe secu i y wi h g aph-based LSTM in log analysis',
Chaos Jou nal. A ailable a : h ps://de gipa k.o g. /en/pub/chaos/a icle/1348302 [Accessed 14 Ma ch 2025].
[12] Alde man, J., Fa ley, N. and C amp on, J. (2017) 'T ee-based c yp og aphic access con ol', in Lec u e No es in
Compu e Science. Sp inge . A ailable a : h ps://link.sp inge .com/chap e /10.1007/978-3-319-66402-6_5
(Accessed: 12 Ma ch 2025).
[13] Ale izos, L., Eiza, M.H., and Ta, V.T. (2022) 'Blockchain-enabled in usion de ec ion and p e en ion sys em o
APTs wi hin ze o us a chi ec u e', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /9862967/ (Accessed: 12 Ma ch 2025).
[14] Alja ah, S.J., Che bal, S., and Mashaleh, A. (2024) 'On he Compa a i e Analysis o T ends in Cybe secu i y Risk
Assessmen , Go e nance, and Compliance F amewo ks', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10847280/ (Accessed: 12 Ma ch 2025).
[15] Anandha aj, N. (2024) 'AI-Powe ed Cloud Secu i y: A S udy on he In eg a ion o A i icial In elligence and
Machine Lea ning o Imp o ed Th ea De ec ion and P e en ion', Resea chGa e. A ailable a :
h ps://www. esea chga e.ne /p o ile/Jou nal-Pub/publica ion/382527337_AI-
Powe ed_Cloud_Secu i y_A_S udy_on_ he_In eg a ion_o _A i icial_In elligence_and_Machine_Lea ning_ o _Imp
o ed_Th ea _De ec ion_and_P e en ion/links/66a1c36127b00e0ca43e65d1/AI-Powe ed-Cloud-Secu i y-A-
S udy-on- he-In eg a ion-o -A i icial-In elligence-and-Machine-Lea ning- o -Imp o ed-Th ea -De ec ion-and-
P e en ion.pd (Accessed: 12 Ma ch 2025).
[16] Ankalaki, S., Rajesh, A.A., Palla i, M., and Hukke i, G.S. (2025) 'Cybe A ack P edic ion: F om T adi ional Machine
Lea ning o Gene a i e A i icial In elligence', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10909100/ (Accessed: 12 Ma ch 2025).
[17] A huko ale, N.K., Yi, C.J., Loh Zi Xin, A.Y., Qi, C.J. and Yu, D.Z. (2025) 'E alua ing Ad anced Cybe secu i y
Technologies o Cloud En i onmen s', P ep in s. A ailable a :
h ps://www.p ep in s.o g/ on end/manusc ip /c7a 352449b61cc7392930a6887c269c/download_pub
[Accessed 14 Ma ch 2025].
[18] Bha acha ya, P., Sa aswa , D., Da e, A., and Acha ya, M. (2021) 'Coali ion o 6G and blockchain in AR/VR space:
Challenges and u u e di ec ions', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /9656726/ (Accessed: 12 Ma ch 2025).
[19] B a sas, C., Anas asiadis, E.K., and Angelidis, A.K. (2024) 'Knowledge G aphs and Seman ic Web Tools in Cybe
Th ea In elligence: A Sys ema ic Li e a u e Re iew', Jou nal o Cybe secu i y and P i acy, 4(3), A icle 25. DOI:
10.3390/jcp4030025 (Accessed: 12 Ma ch 2025).
[20] Casas, P., Vane io, J., Ull ich, J. and Find ik, M. (2023) GRAPHSEC: Ad ancing he Applica ion o AI/ML o Ne wo k
Secu i y Th ough G aph Neu al Ne wo ks. Sp inge . A ailable a :
h ps://books.google.com/books?hl=en&l =&id=5CnKEAAAQBAJ&oi= nd&pg=PA56&dq=g aph-
based+secu i y+models+ o +ai-
d i en+da a+s o age+and+cybe secu i y&o s=Soi3RKBMQS&sig=yauZm0sZ628u23ZX qb _xjMCnk (Accessed:
12 Ma ch 2025).
[21] Cas o, O.E.L., Deng, X., and Pa k, J.H. (2023) 'Comp ehensi e su ey on AI-based echnologies o enhancing IoT
p i acy and secu i y: T ends, challenges, and solu ions', HCISJ. A ailable a :
h p://hcisj.com/da a/ ile/a icle/2023080005/13-39.pd (Accessed: 12 Ma ch 2025).
[22] Chen, C., Zhang, X., Qiu, H., Lou, J., Liu, Z. and Chen, X. (2024) 'MaskA mo : Con idence masking-based de ense
mechanism o GNN agains MIA', In o ma ion Sciences. A ailable a :
h ps://www.sciencedi ec .com/science/a icle/pii/S0020025524004924 [Accessed 14 Ma ch 2025].
[23] Chen, Y., Hu, T., Lou, F., Yin, M., Zeng, T., Wu, G. and Wang, H. (2024) 'A Knowledge G aph-Based Consis ency
De ec ion Me hod o Ne wo k Secu i y Policies', Applied Sciences, 14(18), A icle 8415. DOI:
10.3390/app14188415 (Accessed: 12 Ma ch 2025).
[24] De Alwis, C., Pham, Q.V., and Liyanage, M. (2022) 6G F on ie s: Towa ds Fu u e Wi eless Sys ems. A ailable a :
h ps://books.google.com/books?hl=en&l =&id=7LKeEAAAQBAJ&oi= nd&pg=PR15&dq=gbsm+in eg a ion+wi
h+blockchain-
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1123
[122] S u ano, V.C. (2024) 'Esplo a e le Capaci à dei La ge Language Models nell'O imizza e le Ope azioni della Supply
Chain', Uni e si y o Genoa. A ailable a : h p://uni e.unige.i /handle/123456789/8294 (Accessed: 12 Ma ch
2025).
[123] Syed, N.F., Shah, S.W., Shaghaghi, A., and Anwa , A. (2022) 'Ze o us a chi ec u e (ZTA): A comp ehensi e
su ey', IEEE Xplo e. A ailable a : h ps://ieeexplo e.ieee.o g/abs ac /documen /9773102/ (Accessed: 12
Ma ch 2025).
[124] Tan, Y., Zhang, Q., Li, Y., and Yu, X. (2024) 'AI-D i en Ne wo k Secu i y and P i acy', Elec onics, 13(12), A icle
2311. DOI: 10.3390/elec onics13122311 (Accessed: 12 Ma ch 2025).
[125] Tan, Z., Pa amba h, S. P., Anagnos opoulos, C., Singe , J., & Ma ne ides, A. K. (2025). Ad anced Pe sis en Th ea s
Based on Supply Chain Vulne abili ies: Challenges, Solu ions, and Fu u e Di ec ions. IEEE In e ne o Things
Jou nal, 12(6), 6371–6395. h ps://doi.o g/10.1109/jio .2025.3528744
[126] Ta a da , R. (2024) AI-Powe ed Cybe secu i y Th ea De ec ion in Cloud En i onmen s. Resea chGa e. A ailable
a : h ps://www. esea chga e.ne /p o ile/Resea ch-Schola -Ii/publica ion/389204120_AI-
POWERED_CYBERSECURITY_THREAT_DETECTION_IN_CLOUD_ENVIRONMENTS/links/67b86523461 b56424
e4e379/AI-POWERED-CYBERSECURITY-THREAT-DETECTION-IN-CLOUD-ENVIRONMENTS.pd (Accessed: 12
Ma ch 2025).
[127] Thanalakshmi, P., Ani ha, R., Anbazhagan, N., and Cho, W. (2021) 'A hash-based quan um- esis an chameleon
signa u e scheme', Senso s, 21(24), A icle 8417. DOI: 10.3390/s21248417 (Accessed: 12 Ma ch 2025).
[128] Tsoulias, K., Palaiok assas, G., F agkos, G. and Li ke, A. (2020) 'A g aph model-based blockchain implemen a ion
o inc easing pe o mance and secu i y in decen alized ledge sys ems', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /9130718/ (Accessed: 12 Ma ch 2025).
[129] Ullah, B., Kamal, A., and Asi , S.A. (2024) 'Le e aging A i icial In elligence o Ad anced Cloud Secu i y:
Discussing Techniques and Applica ions', Co al Publica ions. DOI: 10.52633/4yjk 74 (Accessed: 12 Ma ch
2025).
[130] Wajid, S. and Sans, M. (2024) 'In e ne o Things Secu i y: Le e aging AI and Machine Lea ning o Anomaly
De ec ion', Resea chGa e. A ailable a : h ps://www. esea chga e.ne /p o ile/Ma a-Sans-
2/publica ion/389350602_In e ne _o _Things_Secu i y_Le e aging_AI_and_Machine_Lea ning_ o _Anomaly_De
ec ion/links/67b 26238311ce680c75 925/In e ne -o -Things-Secu i y-Le e aging-AI-and-Machine-Lea ning-
o -Anomaly-De ec ion.pd (Accessed: 12 Ma ch 2025).
[131] Wan, C., Wang, Y., Xu, J., Wu, J., Zhang, T., and Wang, Y. (2024) 'Resea ch on p i acy p o ec ion in ede a ed
lea ning combining dis illa ion de ense and blockchain', Elec onics, 13(4), A icle 679. DOI:
10.3390/elec onics13040679 (Accessed: 12 Ma ch 2025).
[132] Wen, S.F., Shukla, A., and Ka , B. (2025) 'A i icial in elligence o sys em secu i y assu ance: A sys ema ic
li e a u e e iew', Sp inge Link. A ailable a : h ps://link.sp inge .com/a icle/10.1007/s10207-024-00959-0
(Accessed: 12 Ma ch 2025).
[133] Wijenayake, D.S. and Henna, S. (2023) 'A G aph Neu al Ne wo k-based Secu i y Pos u e-awa e Cloud Se ice
P o ide Selec ion o Mul i-cloud', IEEE Xplo e Con e ence P oceedings. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10470882/ [Accessed 14 Ma ch 2025].
[134] Wijenayake, D.S. and Henna, S. (2023) 'A G aph Neu al Ne wo k-based Secu i y Pos u e-awa e Cloud Se ice
P o ide Selec ion o Mul i-cloud', IEEE Xplo e Con e ence P oceedings. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10470882/ [Accessed 14 Ma ch 2025].
[135] Win, T.Y., Tian ield, H., and Mai , Q. (2017) 'Big da a-based secu i y analy ics o p o ec ing i ualized
in as uc u es in cloud compu ing', IEEE Xplo e. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /7949076/ (Accessed: 12 Ma ch 2025).
[136] Xin, R., Wang, J., Chen, P. and Zhao, Z. (2025) 'T us wo hy AI-based Pe o mance Diagnosis Sys ems o Cloud
Applica ions: A Re iew', ACM Compu ing Su eys. DOI: 10.1145/3701740 [Accessed 14 Ma ch 2025].
[137] Xu, R., Lan, Q., Pokh el, S.R., and Li, G. (2023) 'A knowledge g aph-based su ey on dis ibu ed ledge echnology
o IoT e icals', ACM Digi al Lib a y. DOI: 10.1145/3609503 (Accessed: 12 Ma ch 2025).
[138] Xun, A.T., En, L.A.Z., Shen, L.T., Xin, A.N., Soon, W.H., and Jun, W.Z. (2025) 'Building T us in Cloud Compu ing:
S a egies o Resilien Secu i y', P ep in s.o g. A ailable a :
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1108-1124
1124
h ps://www.p ep in s.o g/ on end/manusc ip /e07486 c39a494 2a9 999 253da8a79/download_pub
(Accessed: 12 Ma ch 2025).
[139] Ye, M., Ko eshwa a, S., Dunn, D. and F anke, H. (2024) 'Posi ion Pape : F om Con iden ial Compu ing o Ze o
T us , Come Along o he (Bumpy?) Ride', ACM Compu ing Su eys. DOI: 10.1145/3696843.3696848 [Accessed
14 Ma ch 2025].
[140] Ye, W., Qian, C., An, X. and Yan, X. (2023) 'Ad ancing ede a ed lea ning in 6G: A us ed a chi ec u e wi h g aph-
based analysis', IEEE Xplo e Con e ence P oceedings. A ailable a :
h ps://ieeexplo e.ieee.o g/abs ac /documen /10436772/ [Accessed 14 Ma ch 2025].
[141] Zacha is, A., Ka os, V., and Pa sakis, C. (2024) 'In eg a ing AI-d i en h ea in elligence and o ecas ing in he
cybe secu i y exe cise con en gene a ion li ecycle', Sp inge Link. A ailable a :
h ps://link.sp inge .com/a icle/10.1007/s10207-024-00860-w (Accessed: 12 Ma ch 2025).
[142] Zhang, C., Wang, N., Hou, Y.T., and Lou, W. (2024) 'Machine Lea ning-Based In usion De ec ion Sys ems:
Capabili ies, Me hodologies, and Open Resea ch Challenges', TechRxi . A ailable a :
h p://www. ech xi .o g/doi/ ull/10.36227/ ech xi .173627464.48290242/ 1 (Accessed: 12 Ma ch 2025).
[143] Zhang, K. (2023) 'Towa ds Da a P i acy and U ili y in he Applica ions o G aph Neu al Ne wo ks', Geo gia S a e
Uni e si y Disse a ions. DOI: 10.57709/36369498 [Accessed 14 Ma ch 2025].
[144] Zhong, H., Yang, D., Shi, S., Wei, L. and Wang, Y. (2024) 'F om da a o insigh s: he applica ion and challenges o
knowledge g aphs in in elligen audi ', Sp inge Jou nal o Cloud Compu ing. A ailable a :
h ps://link.sp inge .com/a icle/10.1186/s13677-024-00674-0 [Accessed 14 Ma ch 2025].
[145] Zhong, H., Yang, D., Shi, S., Wei, L. and Wang, Y. (2024) 'F om da a o insigh s: he applica ion and challenges o
knowledge g aphs in in elligen audi ', Jou nal o Cloud Compu ing, Sp inge . A ailable a :
h ps://link.sp inge .com/a icle/10.1186/s13677-024-00674-0
[146] Zhuwankinyu, E., Moyo, S., Chi asa, C., & Ncube, S. (2023). E-Wild Li e Ale : Tackling he Human-Wildli e Con lic
P oblem E-Wild Li e Ale : Tackling he Human-Wildli e Con lic P oblem.
h ps://digi alcommons.kennesaw.edu/cgi/ iewcon en .cgi?a icle=1120&con ex =acis
