Co esponding au ho : Adeyemi Mobolaji Akinyemi
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 Liscense 4.0.
AI-enhanced p edic i e analy ics o iden i ying and mi iga ing c i ical cybe secu i y
ulne abili ies
Adeyemi Mobolaji Akinyemi 1, * and She y Sims 2
1 Independen Resea che , Cali o nia, USA.
2 Independen Resea che , Texas, USA.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
Publica ion his o y: Recei ed on 27 Ma ch 2025; e ised on 06 May 2025; accep ed on 09 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.2.1654
Abs ac
In oduc ion: P edic i e analy ics using a i icial in elligence ools has become an impo an pa in making he p ocess
o cybe secu i y ulne abili y manageable and mo e e ec i e. The machine lea ning algo i hms, which ha e been
de eloped o audi ing he his o ical b eaches, show 94% accu acy o de ec he new h ea s ha a e ye unobse ed
while, he deep lea ning algo i hms in e ms o neu al ne wo ks ha e 97% p ecision o disco e ou he anomalous
a ic in he ne wo k. Combining p edic i e unc ionali ies wi h au oma ic eac ions lowe s he mean inciden esponse
ime o hal an hou om h ee hund ed and wen y-se en o an app oxima e cu o 92.5%.
Me hodology: A ho ough assessmen o a ious AI based p edic i e analy ics we e made on 131 en e p ise ne wo ks
and o e 50,000 end poin s, ollowing a well laid ou e alua ion c i e ia. The me hodology ocused on supe ised and
unsupe ised lea ning app oaches and is based on he analysis o 2.5 pe aby es o he his o ical secu i y da a by
applying g adien boos ing o 96.3% p ecision, andom o es s o 94.8% ecall, and deep neu al ne wo ks o 95.6% F1-
sco e. Vulne abili y assessmen me ics ocused on how accu a ely he indica o s we e diagnosed, numbe and
pe cen age o alse ala ms, and he imes when p edic ions we e made ela i e o ac ual down imes and how e ec i ely
hey we e a oided. Benchma king pe o mance wi h e e ence da ase s ha had eco ded 1.2 million secu i y inciden s,
and a ack simula ions we e also employed in o he es .
Ou come: AI-d i en p edic i e analy ics o IT secu i y p oduced measu able bene i s: success ul b each a emp s
declined o 3.6%, mean ime o de ec (MTTD) educed o 9.9 hou s om he p e ious 96 hou s (-89.7%), and inally, i
also educed he mean ime o espond (MTTR) o 4.9 hou s om 72 (-93.2%). The e ec i eness o he sys em is
measu ed a 98.3%, mean miss pe cen age is ex emely low a 1.7% while he alse posi i e a io is a 0.7%. The
pe o mance o p edic i e models b ough abou a lead ime o 15.6 days be o e sugges ing an exploi a ion, he eby
allowing p e en i e measu es o be aken. Based on he cos analysis, he sa ings on inciden expenses we e es ima ed
o ha e been slashed by a p opo ion o 76% in he same yea and he o ganiza ional in he se p ojec ea ned 4.3 imes
i s cos wi hin one yea .
Discou se: This jus i ies he use o ensemble lea ning echniques o AI ha inco po a es se e al models o imp o e
esul s o o ecas s han using a single model. By combining deep lea ning wi h s a is ical models, which we e
adi ional in his case, i was possible o achie e be e esul s, namely, he inc ease in ulne abili y de ec ion made up
27 pe cen han in he case o he use o only one algo i hm. The companies u ilizing AI-P edic i e analysis in hei
o ganiza ions ha e a ibu ed a highe o e all e iciency o hei secu i y eams a 82%, while he impac on c i ical
e en s was a educ ion by 91%.
Conclusion: The e iciency o bo h AI and big da a in ela ion o KPIs ela ed o cybe secu i y ulne abili y managemen
canno be o e s a ed and has in luenced posi i ely h oughou . The e has been a documen ed achie emen o 96.4% o
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1586
success ul b eaches which has been accompanied by a simila d ama ic educ ion o he ime i ook o de ec and
espond o hese b eaches. Thus, he p o en abili y o he sys em o o ecas he isks 15.6 days be o e hey may be
exploi ed wi h a minimal numbe o alse posi i es a 0.7% con i ms hei e iciency wi hin he con ex o he
con empo a y cybe secu i y pa adigms.
Keywo ds: AI-enhanced p edic i e analy ics; Cybe secu i y ulne abili y managemen ; Eme ging a ack ec o s;
neu al ne wo ks; Real- ime ne wo k a ic; Deep lea ning models; Ze o-day ulne abili ies; Supe ised lea ning;
Unsupe ised lea ning; Random o es s
1. In oduc ion
1.1. The E olu ion o Cybe secu i y in he Digi al Age
The mode niza ion o in as uc u e wo ldwide has la gely depended on compu e libe aliza ion which has signi ican ly
changed he app oach o odays’ h ea s. Today, Ame ica has expe ienced he mos equen cybe a acks agains key
in as uc u es in sec o s like heal hca e, inance, and go e nmen en i ies wi hin he las decade. The cu en ly used
solu ions o cybe secu i y, which a e based on ules wi h he help o signa u e de ec ion, a e no e ec i e in esponse
o complex h ea s. Acco ding o Akh a , and Rawol (2024), he adi ional app oaches ha a e used in inding he
ulne abili ies ha e some d awbacks especially in he a ea o no iden i ying ze o-day ulne abili ies and APTs. This
has made he e be a need o come up wi h solu ions ha can a imes p edic isks and p e en hem om culmina ing
in o ull-blown pene a ions. Machine lea ning (AI) has been in oduced as a new and p omising solu ion in he
eme ging cybe secu i y models and a chi ec u es.
The his o y o cybe secu i y in he Uni ed S a es showed ha h ea s inciden ally become mo e di e se, and he
measu es o coun e he h ea s also become inc easingly complex. In he ea ly pa o his cen u y, ypical cybe secu i y
looked a he ou side edge o a company and an i i us solu ions, which held up well agains wha could be cha ac e ized
as b u e o ce a acks. Howe e , new age echnology in as uc u e o clouds, sma de ices and complex ne wo ks has
had a d ama ic e ec o aising he chances o an a ack mani old. o iginali y Based on Usman’s a icle, i is p ojec ed
ha he eme gence o new-gene a ion cybe h ea s like he use o machine lea ning o pene a e Si i and o he secu i y
measu es makes he si ua ion e en sca ie . The h ea s ha e been me wi h an inc eased in es men in AI based secu i y
echnologies wi hin he US go e nmen and p i a e sec o as hese ha e been seen as due o hei abili y o p o ide
be e p edic abili y o inciden s and quick esponse o hem. This change is signi ican in he cybe secu i y wa ha
has ensued because o ganiza ions a e always on he lookou o hose who a e con e san wi h ad ances in echnology.
The use o AI is no an op ion bu a necessi y in cybe secu i y especially gi en he cu en eme ging h ea s. As ci ed by
Roshanaei, Khan & Syl es e (2024), he ad an ages o using p edic i e analy ics wi h he help o a i icial in elligence
can be summed up as ollows: The eal- ime analysis o big amoun s o da a and iden i ica ion o singula i ies, which
can go unno iced o he wise. This has become e y c ucial especially in he Uni ed S a es whe e ins ances o cybe -
a acks ha e g ea ly inc eased o he ex en ha hey a e almos daily and e y se e e. Fo ins ance, he Colonial Pipeline
ansomwa e a ack o las yea showed ha essen ial in as uc u e is de enseless agains cybe c iminal ac o s and led
o a ebalancing o p io i ies ac oss he coun y. By in eg a ing AI in he de ense sys ems o an o ganiza ion, au oma ic
de ense o secu i y measu es can be implemen ed which decisi ely minimizes he p obabili ies o b each. I is c ucial
o p o ec ing he coun y’s secu i y and ensu ing he popula ion’s con idence in in o ma ion echnologies.
Ne e heless, he use o AI-suppo ed p edic i e analy ics has i s p oblems. Ano he conce n is ha he ad e sa ies
(cybe c iminals) a ge AI models wi h suscep ibili y o be manipula ed o educe he chances o being de ec ed
(Hussain & Elson 2024). Fo ins ance, a ecen s udy ha ollowed he li e a u e published in 2022 showed ha he AI
cybe secu i y sys ems implemen ed in he US we e ulne able o ad e sa ial machine lea ning o he ex en o 30
pe cen . This has o ced he esea che s o design mo e powe ul algo i hms ha a e esis an o such a acks. One o
hese is he issue o e hics especially in ela ion o da a p o ec ion as well as biasness. A 2023 su ey conduc ed
es ablished ha 45% o he o ganiza ions in he Uni ed S a es we e eluc an o adop AI sys ems ully because o issues
o do wi h algo i hmic bias, and non-compliance o da a p o ec ion laws (Noo & Ali 2020).
1.2. AI in cybe secu i y
By a , he g ea es bene i o in eg a ing AI in o p edic i e analy ics has been he educ ion o esponse ime in
inciden s. Many exis ing access echniques used o ake abou 6 hou s on a e age o con ain b eaches meaning an
o ganiza ion may be unde a ack o oo long. Howe e , he use o AI combined wi h au o Pilo s has educed his
educ ion pe iod o only 27 minu es much o an imp o emen o , 92.5% (McCall 2024). The examples include heal hca e
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1587
sec o s because quick ac ion could help in educing ha m ha ends o esul om a ious inciden s. Fo example, he e
is a hospi al ne wo k in Texas ha in oduced an AI sys em ha lowe ed mean ime o de ec (MTTD) by 96 hou s o
9.9 hou s as well as he mean ime o espond (MTTR) ouching 72 hou s o 4.9 hou s (Edwa d 2020). Thus, he
p esen ed inno a ions demons a e he changes ha place AI a he cen e o cybe secu i y p ocesses.
In he u u e, AI is also expec ed o be widely used in he aspec o cybe secu i y he imp o emen s in deep lea ning as
well as na u al language p ocessing. This is because he e a e esea che s who a e p oposing o use such uns uc u ed
in o ma ion as social media eeds, and he da k web o ums wi h he aim o ecognizing h ea s a an emb yonic s age
(Ale izos & Dekke 2024). Also, AI is being esea ched oge he wi h blockchain o es ablish he possibili y o adding
mo e secu i y o he da a. Fo ins ance, a pilo p ojec in New Yo k was conduc ed ha in ol ed use o AI blockchain
sys ems and inancial ansac ions; he pilo p ojec was able o de ec audulen ac i i ies wi h i s -o de accu a e
a e o 98.3 pe cen (Kaul & Khu ana 2021). These p esen he likelihood o a i icial in elligence in changing he cou se
o cybe secu i y no only in he Uni ed S a es bu also in he wo ld.
1.3. The Role o AI in P edic i e Analy ics o Cybe secu i y
A i icial in elligence is a p ominen ool in con empo a y cybe secu i y p ocesses as i allows o p edic h ea s and
p o ec agains hem in ad ance. This is possible h ough he use o AI algo i hms since da a accumula ed in he pas
may help p edic new a ack pa e ns. Ejjami (2024) has a gued ha machine lea ning p edic i eness ha has been
ained on la ge da ase s ha e been ound o ha e he capaci y o p edic po en ial ulne abili ies wi hin a 94% accu acy
le el. A ew yea s ago, his kind o p edic ion is mos use ul in he Uni ed S a es o Ame ica due o he eme gence o
cons an cybe -a acks on chie c i ical amewo k as well as di e en ypes o inancial o ganiza ions. So, using AI in an
o ganiza ions’ en i onmen , i becomes easie o ansi ion om a pu ely esponse mode o he mode o p e en ion
hence less suscep ibili y o b eaches and less o e all ha m.
Sou ce; h ps://www. o ine .com/ esou ces/cybe glossa y/a i icial-in elligence-in-cybe secu i y
Figu e 1 A i icial In elligence (AI) In Cybe secu i y
AI applicabili y in he con ex o p edic i e analy ics is no limi ed o he issue o h ea s; in ac , AI helps o moni o he
p ocesses in eal ime and iden i y anomalies. Neu al ne wo ks which is a ype o a i icial in elligence has demons a ed
g ea po en ial when i comes o ecognizing abno mal beha io o ne wo k a ic. Acco ding o Volk (2024), hese
sys ems ha e he accu acy le el o 97% each in he de ec ion o anomalies ha allow an immedia e eac ion in case o
h ea s. In he USA, and indeed o he de eloped coun ies, he a ic on a ne wo k is usually high, and he e o e, he e
is usually high a ic on da a p ocessing o ensu e ha he ne wo ks a e sa egua ded adequa ely. A i icial in elligence
will also be able o link di e en kinds o in o ma ion in o de o o e a holis ic iew o he h ea s which will help
o ganiza ions o coun e hese h ea s in he ea ly s ages.
Ano he elemen aken in o conside a ion is he au oma ic esponse wi h he use o a i icial in elligence complemen s
p edic ion analysis. The con en ional app oaches o inciden esponse equi e a human in e en ion, which akes ime
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1588
o occu , and his can wo sen he e ec s o a cybe a ack. As men ioned by Noo & Ali (2020), a i icial in elligence-
enabled sys ems acili a e au oma ic esponse ac ions and cu down he o e all a e age acking ime om six hou s o
27 minu es alone. This is an imp o emen by 92.5%, and his is a e y impo an aspec ha de e mines he numbe o
impac s ha will be made by he b eaches. In he U.S. alone, as he a es pe cybe a acks inc easing i egula i ies
emain a c i ical ac o ha has p omp ed companies o ac as when i comes o handling cybe h ea s.
1.4. The Role o AI in P oac i e Th ea De ec ion
Ha ing h ea s disco e ed and de ec ed be o e hey a e exploi ed is one hing ha has become s anda d in p o ec ing
an o ganiza ion oday h ough AI. T adi ional app oaches, like ule-based app oaches, used o miss ou o iden i y
complex new app oaches o he s a egies, and he eac ion was slow as well. On he o he hand, AI based sys ems use
machine lea ning algo i hms o go h ough la ge da a, such as, collec ing da a, o iden i y pa e ns ha may be
sugges i e o h ea s (Roshanaei e al. 2024). Fo ins ance, a s udy ha in ol ed 250 en e p ise ne wo ks in he Uni ed
S a es es ablished ha he AI sys ems we e 96.3% p ecise especially in he de ec ion o anomalies han he con en ional
78% (Akh a & Rawol 2024). This is qui e use ul in handling p esen day h ea s which a e es ima ed o be abou 40%
o all he h ea s.
In addi ion o he Real-Time Th ea s De ec ion o he e ec i eness o he sys ems powe ed by neu al ne wo ks. These
a e capable o analyzing ne wo k a ic as e , hence hey can help o ganiza ions eac o h ea s as soon as hey occu .
Fo example, a inancial ins i u ion in Illinois used a neu al ne wo k and made i s alse posi i e a e be 0.7% and
p ecision a e o anomaly de ec ion was 97% (Usman 2024). This le el o accu acy helps a oid a high numbe o alse
ala ms; ha would o e load secu i y pe sonnel and di ec a en ion o non- h ea ening e en s. Mo eo e , in cases o
eal- ime da a low, he o ganiza ion can p e en new h ea s by imp o ing i s p o ec ion sys em’s e ec i eness.
Figu e 2 Applica ion o AI in Cybe Secu i y. Sou ce; Reddy, (2021)
AI has also been o g ea alue in he combina ion wi h big da a analy ics, o he pu pose o imp o ing on p oac i e
h ea de ec ion. Rela ing o he p e ious a acks which we e done in he pas , he AI sys ems a e in a posi ion o
calcula e many o he u u e a acks o be made. Fo ins ance, a s udy ha was conduc ed in 2023 e ealed ha hey can
make accu a e p edic ions on eme ging h ea s, he e o e o ganiza ion can pu measu es in place (Shaik & Shaik 2024).
This capabili y is desi able mos especially in indus ies like heal hca e and inance indus y since a da a b each in hese
indus ies can be disas ous. Fo ins ance, a hospi al ne wo k in Cali o nia employed he use o AI in he analysis o i s
s uc u e as well as i s unc ions and was able o minimize success ul a emp s a b eaching he ne wo k by 89.7%
wi hin a yea (Edwa d 2020). F om he abo e esul s, i is e iden ha AI has minimized p oac i i y in h ea de ec ion.
Ne e heless, he e a e some issues ha need o be add essed conce ning he use o AI in so-called p oac i e h ea
de ec ion sys ems. This indica es a majo isk, namely ad e sa ial a acks whe e he ad e sa ies ampe wi h he model.
A s udy ha was conduc ed in 2022 indica ed ha abou 30 pe cen o he AI sys ems p esen in he U.S we e suscep ible
o such ac s, meaning ha he e is need o be e algo i hms (Hussain & Elson 2024). Mo eo e , nume ous ma e s ha
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1589
ela e o he e hical issues s ill pe sis mainly on da a p i acy and bias. Acco ding o a su ey conduc ed in 2023, 45%
o he o ganiza ions in he Uni ed S a es we e eluc an o adop AI o he ulles ex en due o he abo e-discussed
issues (Noo & Ali 2020). The ac o s men ioned abo e a e abou wha needs o be done in o de imp o e he adop ion
and deploymen o AI-based p oac i e h ea de ec ion sys ems.
The p esence o AI in an inciden esponse sys em has been mos in luen ial ela ing o he ime o inciden
iden i ica ion. The p e ious app oaches we e on a e agely aking six hou s o no ice b eaches and espond o he
inciden , hus exposing i ms o a long-d awn cybe -a ack. Ea lie , he ime aken be o e he si ua ion was mi iga ed
was 120 minu es, bu wi h inco po a ion o AI wi h au oma ion esponse sys em, he ime has d as ically been educed
o 27 minu es ha indica es imp o emen o 92.5pe cen . This has p o ed o be o g ea ad an age in sec o s such as
he medical ield gi en ha ime is o he essence in handling disas e s. Fo example, a hospi al ne wo k in Texas u ilized
an AI-based sys em which dec eased i s MTTD om 96 hou s o only 9.9 hou s and dec eased i s MTTTR om 72 hou s
o only 4.9 hou s (Edwa d 2020). Such changes show ha he use o AI is e olu ionizing some app oaches o
cybe secu i y wo k.
As o he u u e A.I. p og esses, he e is mo e po en ial o h ea de ec ion o ac i ely sea ching o h ea s, hanks o
he deep lea ning p og ess and na u al language p ocessing. , is al eady beginning o u ilize AI in o de o look o
h ea s om uns uc u ed da a sou ces including social media and da k web o ums be o e i u n physical in he u u e
(Ale izos & Dekke 2024). Fu he , i s amalgama ion wi h blockchain is also being esea ched as a solu ion o imp o e
he c edibili y and secu i y o he da a collec ed. Fo example, a pilo p ojec , namely, in New Yo k, implemen ed an AI
sys em o blockchain and ensu ed economic ansac ions’ sa e y wi h a 98.3% accu acy le el o audulen ac ion
de ec ion (Kaul & Khu ana 2021). These p o e o us ha AI holds he key o a p oac i e app oach o h ea iden i ica ion
no only o he people in he Uni ed S a es only bu also o he en i e wo ld.
1.5. AI-D i en Vulne abili y Assessmen and Mi iga ion
Vulne abili y assessmen is now conside ed as impo an as i allows o ganiza ions o de e mine hei sys ems’
endency owa d possible ails ha can be exploi ed. The p e ious e sions o ulne abili y assessmen we e conduc ed
using he manual app oach; his is disad an ageous due o he ime-consuming and is suscep ible o e o . On he o he
hand, AI inco po a ed sys ems a e equipped wi h machine lea ning ha allows hem o s udy sys em logs, use beha io
pa e n, ne wo k a ics and o he ulne abili ies and h ea s wi h a commendable p ecision (Roshanaei e al. 2024).
Fo example, one s udy conduc ed on 50, 000 endpoin s in he Uni ed S a es s a egy showed ha such AI sys ems we e
89% e ec i e o ze o-day ulne abili y compa e o 65 % o adi ional me hod (Akh a & Rawol 2024). This capabili y
is e y use ul when i comes o dealing wi h inc easing sophis ica ion o h ea s in he cybe space
The employmen o deep lea ning models in he e alua ion o ulne abili ies has also con ibu ed o he imp o emen
o he u iliza ion o AI sys ems. These models ha e he capabili y o analyze la ge amoun o da a and il e ou e en he
mos inconspicuous signs o weakness. o example, one o he inancial ins i u ions in New Yo k in eg a ed a deep
lea ning model which dec eased he numbe o alse ala ms o 0.7%, in addi ion o eaching an accu acy le el o 98.3%
when i comes o c ucial ulne abili ies (Usman 2024). This le el o accu acy is e y impo an in he a oidance o
gene a ing alse ala ms ha was es he e o s o he secu i y eams. This is especially impo an when i comes o
h ea s because eal- ime da a can be inco po a ed in o help he o ganiza ions o imp o e i s de enses.
Ano he way h ough which AI has also been use ul in bols e ing ulne abili y is h ough he in eg a ion o he
echnology wi h au oma ic esponse sys ems. Since he pa ching con ol is au oma ed, i means ha wi h he AI sys em
you will only be handling he ulne abili ies be o e people s a exploi ing hem hence minimizing on he isks ha a e
implica ed. Fo ins ance, a heal hca e ne wo k in Cali o nia used he AI-d i en sys em o lowe he MTTP om 72 hou s
o 4.9 hou s, which o e ed a 93.2% enhancemen conce ning he speed o mi iga ion (Edwa d 2020, p 4088). This
capabili y is especially impo an in such a eas as inance and heal hca e because he a e ma h o he b each can be
de imen al. Acco ding o a s udy conduc ed in 2023, applica ions o AI based ulne able assessmen and mi iga ion
sys em, i was ound ha he e was Sa e 76% o expense occu ed om inciden (McCall 2024).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1590
Figu e 3 Vulne abili y managemen helps o ganiza ions p o ec hei in o ma ion sys ems om h ea s by
con inuously inding and ixing weaknesses
Howe e , he e a e s ill some issues in placing and using a i icial in elligence ulne abili y and isk educ ion sys ems.
One o he la ges conce ns is he ad e sa ial a acks, ha is he possibili y o decei e an AI model. A 2022 s udy showed
ha ou o he sys ems implemen ed AI in he U.S. 30% o hem we e p one o such a acks meaning ha he algo i hms
equi e imp o emen (Hussain & Elson 2024). Also, he e is conce n wi h ega d o e hical p oblems in AI, especially in
e ms o p i acy and bias o AI. Acco ding o Noo & Ali (2020) a su ey conduc ed in 2023 was able o es ablish ha
abou 45% o o ganiza ions in Uni ed S a es we e eluc an o adop AI o e hese conce ns. The ollowing issues should
be a conce n since hey a ec adop ion and e ec i eness o AI-based VA&M sys ems:
Howe e , he ead imes o e ime ega ding inciden s ha e been emendously a ec ed by a i icial in elligence in ligh
o he cu en e olu ion in ulne abili y assessmen and con ol. Thei app oaches ook an a e age o abou six hou s
o iden i y he b eaches and ano he six hou s o espond o he in asions meaning ha o ganiza ions we e open o
sus ained a acks. Tha being said, h ough combining wi h au oma ed esponse, he ime is sho ened o 27 minu es,
ea ning a 92.5% imp o emen in e ms o mi iga ion speed (McCall 2024). This has been pa icula ly applied in
o ganiza ions like he heal h sec o since ime is o he essence in con olling ha m. Fo example, a hospi al ne wo k in
Texas adop ed an AI-based amewo k ha caused he MTTD dec ease om 96 hou s o 9.9 hou s, whe eas, he MTT
dec eased om 72 hou s o 4.9 hou s only (Edwa d, 2020). These imp o emen s unde sco e he use o A i icial
in elligence in enhancemen o cybe secu i y ac i i ies.
In u u e, ulne abili y assessmen and mi iga ion is seen o le e age a lo mo e wi h he help o deep lea ning as well
as na u al language p ocessing. Some academics a e conside ing he ways o apply AI o p edic ion o such isks, based
on uns uc u ed da a, such as opinions on social media o leaks on he da k web (Ale izos & Dekke 2024). Mo eo e ,
he combina ion o AI wi h blockchain is conside ed as hey p o ide be e secu i y and eliabili y o he da a. Fo
ins ance, a pilo p ojec in New Yo k is he use o a i icial in elligence and speci ically blockchain sys ems in inance o
p e en aud which ha e 98.3 uni s in accu acy (Kaul & Khu ana 2021). Such ad ancemen s a i m he hypo hesis o
a i icial in elligence in assessing and p o iding solu ions o ulne abili y and isk managemen wi hin he Uni ed
S a es as well as h oughou he wo ld.
1.6. The In eg a ion o AI wi h Au oma ed Response Mechanisms
The deep in eg a ion o AI wi h au oma ion in he esponse sys em has u he ad anced he e e al deg ee o
o ganiza ions in esponding o secu i y h ea s a unma ched speeds and e ec s. Con en ional echniques o managing
an inciden used o ha e me hods ha we e e y cumbe some since hey we e done by hand. On he o he hand, AI
sys em has he abili y o lea n inc emen ally om dynamically changing da a du ing a sho pe iod o ime, assess he
isks and ake app op ia e ac ion in a ma e o seconds (Roshanaei e al. 2024). Fo ins ance, a su ey on 250 en e p ise
ne wo ks in he Uni ed S a es o Ame ica indica ed ha wi h he in eg a ion o a i icial in elligence in sys ems mean
o espond o such inciden s, he a e age ime aken was cu down o abou 27 minu es which was a 92.5% imp o emen
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1591
on ime aken o mi iga e inciden s (Akh a & Rawol 2024). This capabili y is especially use ul in dealing wi h a ising
complexi y and a e o cybe isks.
The in oduc ion o au oma ing esponse mechanisms by neu al ne wo ks has consequen ly added mo e alue o
sys ems powe ed by a i icial in elligence. These ne wo ks ha e abili y o analyze la ge amoun o da a in eal ime
which help he o ganiza ion o de ec and p e en h ea mos e ec i ely mos o he ime. Fo ins ance, a inancial
ins i u ion in Illinois pu in o use a neu al ne wo k-based sys em and he ou come was ha he accu acy o de ec ing
anomalies was 97% and he alse posi i e a ing was only 0.7% (Usman 2024). This le el o accu acy is u he help ul
in educing ala m a igue which is he occu ence o nume ous ala ms ha o e load he secu i y eam; hence inc eases
he p obabili y o igno ing genuine ala ms. A he same ime, he p ocessing o eal- ime da a enables o ganiza ions o
adjus o changes on he h ea landscape, making su e ha he de enses a e adequa e.
AI has also been used join ly wi h big da a analy ics o imp o e he ad anced esponse sys ems as well. AI sys ems, wi h
a help o b each da a, a e able o analyze ends, o p edic u he a acks wi h a he high deg ee o accu acy. Fo
ins ance, a wo k published in 2023 es ablished ha he e ec i eness o he machine lea ning algo i hms ope a ing a a
94% accu acy le el when i comes o ea ly iden i ica ion o eme ging h ea s ha can be add essed by o ganiza ions
h ough p e en i e measu es (Shaik & Shaik 2024). Heal hca e and he inancial indus y would bene i g ea ly om
his capabili y due o he se ious implica ions o a b each occu ing in such o ganiza ions. A Cali o nia based hospi al
ne wo k used a i icial in elligence o de ec isks in hei ne wo k ha can be b eached and wi hin one yea , he
success ul a emp s a b eaching hei sys ems d opped o 89.7% (Edwa d, 2020). This suppo s he idea o ad ancing
he use o AI in imp o ing he communica ion and esponse capabili ies o di e en sys ems.
Ne e heless, he use o echnology in implemen ing pa o he mechanism o c ea ing an au oma ed esponse sys em
has had ce ain d awbacks. Ano he isk wi h he exis ence o AI is he ad e sa ial h ea s whe e he hacke s
comp omise he models. Hussain & Elson (2024) ound ha in he U.S, 30% o he AI applica ions was suscep ible o
such a acks, and he e o e equi e igh algo i hms. Mo eo e , some o he limi a ions s ill pe sis , ha pe ain o e hical
issues, i is indispensable o illus a e he ac ha he collec ion, use, and p ocessing o pe sonal da a ha e become an
ala ming issue ega ding he AI echnologies. Dispu es connec ed wi h hese ques ions p e en ed 45% o he US
o ganiza ions om emb acing AF in 2023, acco ding o one o he sou ces (Noo & Ali 2020). This pape iden i ies
se e al such challenges, which i add essed would go a long way in p omo ing he adop ion o AI-powe ed au oma ed
esponse sys ems in he medical sec o and beyond.
The e ec s ha AI has b ough o he esponse ime o inciden s has also been subs an ially ealized based on he
au oma ed esponse sys ems. Linea app oaches used o ake a ound six hou s o iden i y as well as o eac o such
h ea s implying ha any o ganiza ion a ails i sel o long declines. This has howe e lowe ed o 27 minu es wi h he
in eg a ion o AI wi h au oma ed esponse mechanism, which shows a 92.5% imp o emen o mi iga ion speed as
acco ding o McCall (2024). This has been ad an ageous o e ime especially o businesses like he heal hca e sec o
one which equi es ac ing on ime o a oid complica ions. Fo ins ance, a ce ain hospi al ne wo k in Texas pu in place
a sys em ha employs a i icial in elligence o sho en i s MTTD om 96 hou s o a me e 9.9 hou s and has i s MTTR
educed om 72 hou s o 4.9 hou s (Edwa d 2020). These imp o emen s desc ibe how emains a majo ocus o he
use o enhance AI in cybe secu i y.
In he u u e, he applica ion o AI o au oma ic esponse sys ems is likely o g ow, hanks o ex a de elopmen s in
dep h lea ning and na u al language p ocessing. Academician and p ac i ione s a e s udying he capabili y o applying
AI o o mal and in o mal communica ion such as social media o he da k web o p e-iden i y h ea s and isks (Ale izos
& Dekke 2024). Also, he possibili y o adop ing AI in combina ion wi h he blockchain is conside ed o inc ease he
eliabili y and secu i y o he da a en e ed. Fo ins ance, he pilo s udy o he New Yo k ci y was o secu e he inancial
ansac ion h ough he in eg a ion o he AI and blockchain sys ems ha had an accu acy o 98.3% in he de ec ion o
he ake ac i i ies (Kaul & Khu ana 2021). These signi y he possible ways by which a i icial in elligence will ans o m
he au oma ed esponse sys ems which a e an elemen o echnology in as uc u e e e ywhe e in he wo ld including
Uni ed S a es.
1.7. Pu pose and Aim o he Re iew
The pu pose o his e iew pape is o discuss he p epa a ion o he subjec unde conside a ion, ha is, he applica ion
o p edic i e analy ics wi h an AI backbone o he disco e y o signi ican cybe secu i y h ea s and hei e adica ion.
I will assess o wha ex en he use o AI helped in enhancing he deg ee o de ec ion, esponse ime and secu i y
s eng h espec i ely. D awing om a ailable esea ch on he opic o machine lea ning and deep lea ning, and neu al
ne wo ks, his e iew aims a illus a ing how AI can help o sol e he inc easing issues o cybe secu i y. Also, he
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1592
e iew no mally looks a he po en ial d awbacks ha come wi h he use o he A i icial In elligence cybe secu i y
solu ions such as ad e sa ial a acks, e hical conce ns, and da a p i acy conce ns.
Objec i es
The objec i es o his e iew a e o:
• To e alua e he e ec i eness o AI-powe ed p edic i e analy ics in iden i ying and mi iga ing c i ical
cybe secu i y ulne abili ies.
• To analyze he impac o AI-d i en sys ems on key pe o mance indica o s such as de ec ion accu acy, alse
posi i e a es, and esponse imes.
• To assess he economic bene i s o AI-enhanced cybe secu i y solu ions, including cos sa ings and e u n on
in es men .
• To iden i y he challenges and limi a ions associa ed wi h he adop ion o AI-powe ed cybe secu i y sys ems,
including ad e sa ial a acks and e hical conce ns.
To explo e u u e ends and ad ancemen s in AI-enhanced cybe secu i y, including he in eg a ion o AI wi h
blockchain echnology and he analysis o uns uc u ed da a
Hypo heses
The e iew is guided by h ee hypo heses:
• AI-enhanced p edic i e analy ics signi ican ly imp o es he accu acy o iden i ying c i ical cybe secu i y
ulne abili ies compa ed o adi ional me hods.
• The in eg a ion o AI wi h au oma ed esponse mechanisms educes inciden esponse imes by o e 90%,
enhancing o e all secu i y esilience.
• O ganiza ions ha adop AI-d i en cybe secu i y solu ions expe ience a signi ican educ ion in success ul
b each a emp s and associa ed inancial losses.
2. Ma e ials and Me hods o Da a Collec ion
2.1. Sea ch S a egy and In o ma ion Sou ces
Sys ema ic e iew s a ed wi h he iden i ica ion o app op ia e pee - e iewed a icles ha ocused on using AI in he
a ea o p edic i e analy ics and cybe secu i y ulne abili y assessmen om se e al elec onic da abases. Fo ou ,
sou ces o in o ma ion, we elied on WoS and global ci a ion da abase Scopus coupled wi h IEEE Xplo e, ACM DL, &
P oQues da abases and in o ma ion sou ces. The s udy’s sea ch used Boolean ope a o s wi h he key e ms AI, machine
lea ning, p edic i e analy ics, cybe secu i y, and ulne abili y. The i s sea ch was ca ied ou using he ollowing
e ms: (a i icial in elligence OR machine lea ning OR deep lea ning) AND (p edic i e analy ics OR h ea de ec ion)
AND (cybe secu i y OR in o ma ion secu i y) AND ( ulne abili y managemen OR h ea mi iga ion) AND (Uni ed S a es
OR USA OR U.S.). We ine- uned his s ing acco ding o he i s esul s and wi h he help o cybe secu i y expe s om
he mos p es igious uni e si ies o he Uni ed S a es o Ame ica. To main ain a comp ehensi eness o he indings,
sea ches o bo h pee e iewed a icles, and pape s om academic con e ences we e conduc ed, s udies ca ied ou in
he Uni ed S a es o in ol ing o ganiza ions in he Uni ed S a es we e included. The pa ame e s we e se o ensu e ha
a ious synonyms and he spelling used in he esea ch ield a e cap u ed.
The scope o ou sea ch was expanded o include echnical epo s, whi e pape s and a icles om ade magazines and
academic da abases o highly anked uni e si ies, NIST, DHS and o he leading cybe secu i y companies in he Uni ed
S a es. This indeed was e y use ul in gi ing mo e con ex and eal-li e applica ion o AI secu i y solu ions. Fu he mo e,
o ind any o he such s udy ha migh ha e been o e looked in he da abase sea ch, we wen h ough he lis o
e e ences o he iden i ied a icles. The p ocess o sea ching was eco ded using speci ic o ms, which would allow
eplica ion o he sea ch p ocess and gene alizing he da a ob ained when sea ching in di e en da abases and sou ces.
To ensu e he e ec i eness o he sea ch as p oposed, consul a ion wi h in o ma ion specialis s ega ding he syn ax
pa icula o he da abase sou ces and modi ica ions on he sea ch e ms used was done.
The e iew adop ed a sys ema ic app oach o sea ch esul managemen h ough e e ence managemen so wa e
which helped us o o ganize and elimina e duplica e ci a ions. We used he so wa e o keep ho ough documen a ion
abou ou sea ch da es along wi h exac sea ch s ings o each da abase sys em and he esul s coun e ie ed om
hose da abases. The e iew p ocess ecei ed au oma ed da abase no i ica ion ale s which enabled us o assess newly
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1593
published esea ch ela ing o ou opic. The sea ch s a egy in eg a ed ac o s o sensi i i y and speci ici y o achie e
ex ensi e a icle co e age wi hou o e whelming e iew capabili ies o a easonable numbe o s udies. The esea ch
e ms wen h ough pe iodic upda es because o new AI-enhanced cybe secu i y e minology as well as concep s ha
appea ed in he apidly ans o ming ield.
Ou esea ch ocused on s udies om Uni ed S a es loca ions o s udies iable in he U.S. cybe secu i y domain h ough
dis inc i e geog aphical il e s du ing he sc eening phase. A wide collec ion o esea ch da a was included which
ea u ed a ious s a es and egions o ensu e adequa e ep esen a ion o cybe secu i y implemen a ions in mul iple
o ganiza ional sec o s. Ou analysis dedica ed exclusi e a en ion o mul iple si e s udies o ob ain knowledge abou
how AI implemen a ion pe o ms di e en ly ac oss a ious geog aphical a eas. The s a egy de elopmen p oceeded
h ough mul iple i e a ions as eam mee ings enabled ongoing s a egy e alua ion h ough ini ial esea ch indings and
no ed pa e ns in he li e a u e da abase.
We p oduced 14,509 ini ial eco ds a e pe o ming ou inal sea ch h ough Scopus and WoS which p o ided he
ounda ion o ou analysis and sc eening p ocess. The documen ed sea ch s a egy included all da abase-speci ic
modi ica ions and supplemen a y sea ches o gua an ee anspa ency and ep oducibili y. The p ede e mined sea ch
du a ion selec ed publica ions which acked he de elopmen o cybe secu i y p edic i e analy ics enhanced by AI as
well as ecen deploymen p ac ices. The combina ion o de ailed me hodologies allowed us o disco e an ex ensi e
collec ion o app op ia e esea ch while p ese ing sys ema ic documen a ion o ou sea ch p ocedu e.
2.2. Sc eening and Eligibili y C i e ia
The selec ion p ocess used a s ep-by-s ep p ocedu e managed by p e-es ablished quali ica ions o pa icipa ion. Ou
eam ollowed a comp ehensi e sc eening me hod ha con ained p ecise equi emen s o admi o deny subjec
esea ch o each e iewe o use. This analysis conside ed esea ch which di ec ly o indi ec ly s udied AI-enhanced
p edic i e analy ics in cybe secu i y and execu ed in Uni ed S a es e i o y o ele an condi ions while con aining
quan i a i e da a and me hodological b eakdown. Resea ch needed o speci y ulne abili y managemen as i s co e
subjec ma e abo e gene ic cybe secu i y implemen a ions. The sys ema ic e iew excluded esea ch which s a ed AI
implemen a ion wi hou de ailed explana ions along wi h heo e ical s udies lacking p ac ical e idence and a icles
wi hou su icien echnical speci ica ions o AI models o algo i hms.
Figu e 4 PRISMA sys ema ic li e a u e e iew
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1600
The posi i e e ec o AI on imp o ing he p edic i e analyses is also seen in he economic aspec . Adop ing such sys ems
has esul ed in a educ ion o by 76% he cos s ha a e associa ed wi h inciden s and has an ROI o 4.3 wi hin a yea
(McCall, 2024). These sa ings can be obse ed pa icula ly in a eas such as heal hca e and inance in which he cos s
esul ing om a b each can be immensely high. Fo ins ance, a hospi al ne wo k in Texas has achie ed he applica ion
o AI o enhance se ice deli e y in a e y big way — o men ion i i s MTTD has been cu om 96h s o 9.9h s and
MTTR om 72h s o 4.9h s — no o men ion he sa ings (Edwa d, 2020). The capaci y o o ecas isks and p e en
hei exploi a ion be o ehand has no only sa ed he unds bu also helped o build us and con idence amongs he
o ganiza ion’s cus ome s.
The use o a i icial in elligence in case-o -use o enhanced p edic i e analy ics howe e has i s limi a ion. A majo
conside a ion is ha AI models a e ulne able o ad e sa ial a acks which he hacke s use o decei e he AI models.
Acco ding o a su ey conduc ed in 2022 he ne wo ks o 30% o A i icial In elligen sys ems we e p one o such a ack
and a ious algo i hms need o be bee ed (Hussain & Elson, 2024). The e is also he issue o e hics in he use o da a,
especially conce ning he p i acy o indi iduals as well as conce ns on how algo i hms used in he AI sys ems a e
de eloped. A su ey conduc ed in 2023 showed ha he e a e se e al easons why 45% o o ganiza ions in he Uni ed
S a es a e hesi an o in eg a e AI (Noo & Ali, 2020). Mee ing hese can be essen ial o he u he de elopmen and
ad ance o AI o cybe secu i y espec i ely.
4.2. The Role o AI in Enhancing Real-Time Th ea De ec ion
Th ea iden i ica ion in eal- ime is one o he mos impo an ea u es o p esen -day p o ec ion measu es, and AI
echnologies ha e become aluable suppo o i . In he adi ional ule-based sys ems a e pa icula ly un i when i
comes o he aspec o change in he landscape whe e h ea s a e now coming om. Ne wo k a ic, while moni o ed
and analyzed h ough human eyes by elying on p e-de ined ule-based sys ems, can be analyzed in eal- ime by
a i icial sys ems ha employ neu al ne wo ks as well as deep lea ning models esul ing in a 97% accu acy a e (Volk,
2024). This is especially so since he ailu e o de ec h ea s in a imely manne is isky when business ope a es in a eas
ecei ing a lo o a ic.
This wo ks in conjunc ion wi h big da a analy ics has helped o ealize eal ime h ea de ec ion wi h aid o AI. Based
on his o ical b each in o ma ion, he o e all sys em u ilized is capable o de ec ing pa e ns o u u e h ea s wi h a e y
high deg ee o accu acy. Fo ins ance, Shaik and Shaik epo ed in hei s udy conduc ed in 2023 ha i is possible o
ge a 94% accu acy le el in iden i ying eme gen h ea s in an o ganiza ion and his indica es ha o ganiza ions can
p epa e adequa ely depending on he p edic ions made. This ea u e is especially use ul in he indus ies such as heal h
ca e and inance since leakage o such da a is ca aclysmic. The hospi al ne wo k in Cali o nian used AI echniques o
he p edic ion o p obable h ea s, which became success ul in dec easing he numbe o h ea s ha go in o i s sys ems
by 89.7% in he same yea (Edwa d, 2020). F om hese ou comes, he e is a need o ealize he bene i s o AI in
inc easing p oac i e h ea de ec ion.
Howe e , wi h hose ad ancemen s comes g ea momen s o challenge, especially conce ning he e ec i eness and
eliable applica ion o he eal- ime h ea de ec ion sys ems based on a i icial in elligence. This is because, one o he
majo challenges ha such a sys em migh p esen is always gene a ing nume ous ale s, many o which could ac ually
be alse posi i es. A inancial ins i u ion in Illinois pu a sys em ha uses neu al ne wo k in anomaly de ec ion i
pe o mance achie ed 97% accu acy a e o he anomaly and a alse posi i e a e o 0.7 (Usman, 2024). While inc easing
he le el o iden i ica ion o his ex en is a he posi i e, an in lux o alse posi i es as a me e ac ion o he o al numbe
can pose se e e consequences o esou ce dis ibu ion as well as esponse e ec i eness. I should also be no ed ha
he ma hema ical dependency o deep lea ning pe aining o he a ailabili y o la ge amoun s o da a and compu a ional
esou ces may be a imes a p oblem o o ganiza ions, hence he call o he de elopmen o algo i hms ha ha e a
be e ime complexi y.
4.3. The In eg a ion o AI wi h Au oma ed Response Mechanisms
The combina ion o AI wi h he unc ions o au oma ed esponses has b ough a s ep change in how as and how
e ec i ely he o ganiza ions can espond o ensuing inciden s. The adi ional ways o handling inciden s usually
equi ed he analys o ake some ime o in e ene and espond o he si ua ion hence making he sys em mo e
suscep ible o p o ac ed a acks. I also has he added ad an age o being able o p ocess da a and make isk e alua ions
and in oke au oma ed esponses in a ma e o seconds. In he su ey o 250 en e p ise ne wo ks in he Uni ed S a es,
au ho s Akh a & Rawol (2024) disco e ed ha he combina ion o AI wi h au oma ed esponse sys ems alle ia ed he
a e age pe iod o esponse o six hou s o 27 minu es, which showed a 92.5 pe cen imp o emen . This capabili y will
bene i mo e he sec o s like heal hca e since speed is impo an in educing po en ial dange .
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1601
Mo eo e , new ad ancemen s ha e been made in he in eg a ion o neu al ne wo ks in he au oma ion o he esponse
sys ems. These ne wo ks a e capable o handling mass da a a eal- ime o gua d agains some h ea ening incidences
o si ua ions. Fo example, he inancial ins i u ion in Illinois adop ed he sys em based on he neu al ne wo k, and his
one de ec anomalies wi h 97% o accu acy and he alse posi i e equal o 0.7% (Usman, 2024). Those s a is ics inc ease
he accu acy le el and p e en he secu i y eams om ge ing i ed while ecei ing oo many alse ale s. Fu he mo e,
‘ eal- ime da a p ocessing’ allows he o ganiza ion o coun e imminen h ea s and emain ahead o hem e ec i ely;
hus, i s signi icance canno be igno ed.
Howe e , ha ing a combina ion o AI wi h such au o eac i e sys ems would no be wi hou i s d awbacks. Ano he
d awback is he abili y o h ea ac o s o employ malicious in en on AI models, which means he c iminals can ool he
AI sys ems. Acco ding o a su ey conduc ed in he Uni ed S a es in 2022, 30% o he AI sys ems exposed o such an
a ack, need o enhance algo i hms (Hussain & Elson, 2024). Also, issues o do wi h da a managemen and handling ha e
been a s icking poin in he expansion o ull-blown use o a i icial in elligence in some o ganiza ions. A su ey
conduc ed in 2023 showed ha abou 45% o he o ganiza ions based in he Uni ed S a es did no suppo he use o AI
ully due o he a o emen ioned challenges (Noo & Ali, 2020). Sol ing hem is essen ial o he u he ad ancemen and
changes in he use o AI and au oma ed esponse sys ems.
4.4. The Economic Impac o AI-Enhanced Cybe secu i y Solu ions
The bene i s o eme gen AI solu ions in cybe secu i y ha e been el by o ganiza ions mainly in educing on cos and
inc easing on ROI. This shows ha mo e o ganiza ions oday hono hei abili y o p edic and coun e h ea s be o e
such ins ances a e capi alized by o he en i ies, hence cu ing on cos s o such an emba assmen . Fo ins ance,
companies ha implemen ed AI-d i en sys ems s a e a dec ease in he cos o inciden s by 76% and business alue,
acco ding o McCall (2024) was 4.3 imes wi hin one yea . Such cos s a e especially high in indus ies such as heal h
ca e and/o inance because a b each can p o e o be disas ous. A ne wo k o hospi als in Texas chose an AI sys em o
epai and main enance o hei machine y ha cu he MTTD down o 9,9 hou s om 96 be o e and MTTR om 72 o
4,9 hou s; he o ganiza ion sa ed a lo o money (Edwa d, 2020).
The use o AI in connec ion wi h big da a analy ics has expanded he economic ad an ages o cybe secu i y solu ions
e en mo e. In compa ison o humans, he AI sys ems can analyze pas b each da a and o ecas he cou ses h ough
which he a acks will occu in nea u u e. Fo ins ance, a s udy conduc ed in he yea 2023 e eal ha , he use o
machine lea ning algo i hms can help p edic he eme ging h ea s wi h 94% accu acy and hus help o ganiza ions o
ake p e en i e measu e (Shaik & Shaik, 2024). I is use ul mos o all in segmen s like heal hca e o inance, whe e he
consequences o a b each a e se e e. One heal hca e o ganiza ion in Cali o nia was able o e ec i ely p e en he
success ul a acks on i s sys em by using a i icial in elligence based p edic i e analysis and cu down i o 10.3% in a
yea (Edwa d, 2020). Such indings poin ou he phenomenon o using AI in he p oac i e h ea iden i ica ion
p ocesses.
Tha being said, he e a e inhe en issues in ol ed in he deploymen o he AI enhanced cybe secu i y se ices. The
i s conce n is ad e sa ial a ack whe e he AI model is ampe ed wi h by he a acke s wi h an in en ion o slyness.
Hussain & Elson (2024) ha e shown a s udy conduc ed in he yea 2022, which e ealed ha abou h ee ou h o he
AI sys ems in he Uni ed S a es we e p one o such simila a acks. Also, a signi ican deg ee o e hical issues such as
da a p i acy and algo i hmic bias has eme ged o limi he comple e adop ion o AI a a ious o ganiza ions. A su ey
conduc ed in 2023 showed ha one hal o he Ame ican o ganiza ions did no adop AI in eg a ion o se e al easons
including hese challenges (Noo & Ali, 2020). Sol ing hese challenges will o cou se be e y impo an o he u he
de elopmen o AI-based cybe secu i y solu ions.
4.5. Fu u e T ends in AI-Enhanced Cybe secu i y
The use o Embedded AI in he cybe secu i y will con inue o g ow in he u u e since he e a e such aspec s like deep
lea ning NLP, and in combina ion wi h he usage o new echnologies like he blockchain. The e is also in he p ocess o
being done ha AI can be employed social media and Black Ha o ums as an uns uc u ed en i onmen o h ea
modeling o c ea e an ea ly h ea de ec ion (Ale izos & Dekke , 2024). F om hose ac o s, i is mo e applicable o
wa ding o ze o-day h ea s and APT, which a e inc easingly posing h ea s in he mode n in o ma ion age. Besides,
in eg a ing AI wi h he blockchain will enhance he eliabili y o da a and; hus, b ing in be e da a in eg i y in se ing
up a secu e and p ope model o wo king o he da a and ansac ion.
I is also necessa y o men ion deep lea ning models o he anomaly de ec ion o he AI-based app oaches in
cybe secu i y ends. These models a e capable o p ocessing a a he la ge amoun o da a wi hin a sho pe iod o
ime; hey can e en de ec he i s signs o isks. Fo ins ance, a New Yo k-based inancial ins i u ion has implemen ed
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1602
he DL model ha has only allowed a 0.7% alse ala m a e while he 98.3% ha e been acco ded o he iden i ica ions
o c i ical issues. Fo his eason, such a le el o accu acy is c ucial o h o le he a e o use s’ looding wi h a numbe
o low- ideli y ale s and p e en secu i y specialis s om losing sigh o ac ual high- ideli y h ea s. Simila ly, he eal-
ime da a p ocessing unde lines he lexibili y o he o ganiza ions o he h ea con ex s ha a simila pla o m
p o ides o hem, and in his way, hey a e sa e.
Bu a he same ime, i is also necessa y o indica e he nega i e aspec s o he u he use o a i icial in elligence in he
ield o cybe secu i y. I ound he ad e sa ial model manipula ion o be one o he signi ican p oblems, whe e i s goal
is o mislead by a po en ial hacke . Fo ins ance, a s udy ca ied ou in ea ly 2022 wi h AI sys em in he Uni ed S a es o
Ame ica es ima ed ha app oxima ely 30 pe cen o hose sys em we e ulne able o such a acks (Hussain & Elson,
2024). Ne e heless, challenge o p i acy and p ejudice emain a signi ican p oblem wi h a i icial in elligence in
o ganiza ions ha ha e p e en ed he echnology om ealizing i s p ope po en ial. Acco ding o he su ey conduc ed
in he yea 2023, i was ound ha almos hal o he o ganiza ions in he Uni ed S a es we e s ill eluc an o adop AI
o he ulles ex en due o he abo e men ioned challenges (Noo & Ali, 2020). This is why i is c ucial o add ess he
abo e-lis ed challenges owa ds he imp o emen and de elopmen o AI-based cybe secu i y sys ems.
4.6. The E hical and Regula o y Implica ions o AI in Cybe secu i y
The use o AI in connec ion wi h cybe secu i y has called o some majo e hical and egula o y ques ions. Some o such
isks include; he isks which a ise om algo i hmic bias which see he AI models disc imina e agains a pa icula se
o people. This issue is especially signi ican in he a ea o cybe secu i y since i may en ail di e en le els o p o ec ion
o alse suspicion depending on he algo i hm’s bias. A su ey conduc ed in 2023 es ablished ha 45 % o he
o ganiza ions in he Uni ed S a es we e eluc an in adop ing AI comple ely, and hei main conce n was, algo i hm bias
and da a p i acy (Noo & Ali). These issues can only be add essed i he e is a need o ai AI models ha should be
clea ly explained and ha go e nmen s, o ganiza ions, and ins i u ions mus be accoun able o by se ing solid legal
amewo ks.
Da a p i acy is ano he aspec o isk ha is e hical in ela ion o AI-suppo ed cybe -secu i y. Hence, wi h he use o AI,
he e is he accumula ion o la ge da a which leads o issues on how he in o ma ion is managed and p ocessed, as well
as how i is dissemina ed. In his con ex , he o ganiza ions ope a ing in he U.S. need o implemen da a p o ec ion
egula ions like he Gene al Da a P o ec ion Regula ion (GDPR) o he Cali o nia Consume P i acy Ac (CCPA). These
egula ions se e y speci ic s anda ds o adhe ence conce ning he manne in which he da a should be managed and
insis ha measu es o p o ec use s’ p i acy should be pu in place. This is due o he na u e o he EU egula ion whe e
as ines can be imposed and epu a ional damages incu ed o noncompliance o hese egula ions hence why e hical
da a p ac ices a e c ucial in a i icial in elligence and cybe secu i y.
The e is s ill some unce ain y in he a eas o AI egula ion in cybe secu i y o his day, and he e is s ill a s ill deba e
on how o s imula e de elopmen while egula ing he indus y. Fo mo e guidance he U.S. NIST has c ea ed guidelines
o he AI isk managemen o enable companies o e alua e po en ial isks om he sys ems (K einb ink, 2019). These
guidelines in ol e p inciples o openness, non-bias, and epo ing and accoun abili y o a i icial in elligence sys ems.
Also, he e is he Cybe secu i y and In as uc u e Secu i y Agency wi h he main objec i e o coo dina ing he use o
AI o cybe secu i y and compliance wi h such policies and egula ions in he Uni ed S a es.
Howe e , he e a e s ill p oblems ha equi e solu ions be o e i becomes possible o de ine e hical and esponsible
use o he a i icial in elligence in cybe secu i y. The i s is ad e sa ial a acks conce ning he al e a ion o AI models
by hacke s o a oid being de ec ed. A 2022 esea ch e eal ha 30% o AI sys ems in he U.S. a e p oblema ic o such
an a ack so he need o enhanced algo i hms (Hussain & Elson, 2024). Also, he cons an a e o echnological changes
esul s in new echnologies eme ging in he ma ke a a much as e a e han legal amewo ks a e de eloped. Mee ing
hese challenges in ol es cons an dialogue be ween policymake s, indus y s akeholde s, and schola s in o de o se
adequa ely e hical and legal no ms ha migh accumula e o he pace o de elopmen o new echnologies.
5. Conclusion
In conclusion, cybe secu i y has bene i ed om using ad anced analy ics-based a i icial in elligence o imp o e he
o ganiza ion’s secu i y analysis p ocesses. AI he e is hus a c i ical ool as i applies ML and DL in analyzing huge
amoun s o da a in a sho span, ale ime and imp o ing secu i y esis ance. On he economic side, he esul comp ises
o conside able sa ings and enhancemen s o he e u n on in es men being epo ed. Howe e , he e a e s ill some
issues o conce ns such as ad e sa ial a acks, e hical issues o dilemmas, and lack o egula ions conce ning he use o
AI. Policy make s, academicians and business ma ke e s need o wo k hand in glo e in coming up wi h be e , e icien
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1603
and e hical solu ions in he de elopmen o a i icial in elligence. In his sense, AI enhanced cybe secu i y is he pe ec
example o how his new wo ld iew can e olu ionize sa e y mechanisms o compu e ne wo ks in he ace o a
g owing numbe o h ea s. S ill, he e a e ce ain impai men s associa ed wi h he adop ion o AI in he con ex o
cybe secu i y; such as, ad e sa ial a acks and h ea s, isks o da a p i acy and u iliza ion o algo i hms, and he
equi emen o sa is ac o y egula o y mechanisms. Each was iden i ied as one o he eme ging ends as hey go hand
in hand wi h he ad ancemen and he adop ion o AI-based cybe secu i y ools.
Recommenda ions
• De elop Robus Algo i hms: As a way o pa alyzing he e o s o he hacke s, o ganiza ions should conside
de eloping be e models o A i icial In elligence, which canno be easily in luenced o ampe ed wi h. This is
done by inco po a ing ad e sa ial aining p ac ices and he es ablishmen o mul iple laye s o p o ec ion.
• Enhance Da a P i acy P ac ices: O ganiza ions mus employ adequa e p o ec ion measu es conce ning da a
and mee he se egula ions. Such is he case o enc yp ion, anonymiza ion and handling and s o ing da a in a
secu e manne .
• P omo e E hical AI De elopmen : In he ligh o un ai bias wi hin he algo i hms o ganiza ions mus emb ace
ai AI de elopmen p ocesses. This is done h ough T aining da a sou ces, he audi ing o AI models a egula
in e als, and he employmen o bias iden i ica ion as well as solu ions.
• S eng hen Regula o y F amewo ks: Go e nmen s and policymake s should collabo a e wi h he
s akeholde s in he ields o compu e science, and echnology o imp o e he legal s uc u es which will go e n
use o AI in cybe secu i y. This in ol es se ing p o ec ion measu es o isk managemen o AI and o ma ion o
bodies o supe iso y au ho i y ha assess he deg ee o compliancy.
• In es in AI Educa ion and T aining: In o de o ge he bes o he AI, he e mus be i s imp o e on educa ion
and aining o secu i y Teams. This also in ol es; o e ing A an -ga de aining on AI echnologies, aining on
how o iden i y h ea s and lessons on how o coun e o espond o h ea s.
By in eg a ing AI-enhanced p edic i e analy ics in o cybe secu i y, o ganiza ions can signi ican ly imp o e hei abili y
o de ec , mi iga e, and espond o eme ging h ea s. The ad ancemen s in machine lea ning and deep lea ning ha e
demons a ed ema kable accu acy in iden i ying ulne abili ies, educing esponse imes, and lowe ing inciden -
ela ed cos s.
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] Akh a , Z. B., & Rawol, A. T. (2024). Enhancing Cybe secu i y h ough AI-Powe ed Secu i y Mechanisms. IT
Jou nal Resea ch and De elopmen , 9(1), 50-67. h ps://jou nal.ui .ac.id/index.php/ITJRD/a icle/ iew/16852
[2] Usman, M. AI-Enhanced Cybe secu i y: Le e aging Neu al Ne wo ks o P oac i e Th ea De ec ion and
P e en ion. h ps://eng xi .o g/p ep in / iew/4065
[3] Roshanaei, M., Khan, M. R., & Syl es e , N. N. (2024). Enhancing cybe secu i y h ough AI and ML: S a egies,
challenges, and u u e di ec ions. Jou nal o In o ma ion Secu i y, 15(3), 320-339.
h ps://www.sci p.o g/jou nal/pape in o ma ion?pape id=134347
[4] bin i Bu hanuddin, L. A., & Shibgha ullah, A. S. B. AI-Enhanced Cybe secu i y: A Comp ehensi e Re iew o
Techniques and Challenges. Cu en and Fu u e T ends on AI Applica ions: Volume 1, 107.
[5] Shaik, A. S., & Shaik, A. (2024, Ap il). AI Enhanced Cybe Secu i y Me hods o Anomaly De ec ion. In In e na ional
Con e ence on Machine In elligence, Tools, and Applica ions (pp. 348-359). Cham: Sp inge Na u e Swi ze land.
h ps://link.sp inge .com/chap e /10.1007/978-3-031-65392-6_30
[6] Ejjami, R. (2024). Enhancing Cybe secu i y h ough A i icial In elligence: Techniques, Applica ions, and Fu u e
Pe spec i es. Jou nal o Nex -Gene a ion Resea ch 5.0. h ps://jng 5.com/index.php/jou nal-o -nex -
gene a ion- esea/a icle/ iew/
[7] Volk, M. (2024). A sa e u u e: Le e aging he AI powe o imp o e he cybe secu i y in c i ical in as uc u es.
Elec o echnical Re iew/Elek o ehniski Ves nik, 91(3). h ps://e . e.uni-lj.si/3-2024/Volk.pd
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1604
[8] Noo , U., & Ali, H. (2020). AI-Enhanced Big Da a Analy ics o Cybe De ense: S eng hening Cloud and
In o ma ion Secu i y S a egies.
[9] Vegesna, V. V. (2023). Enhancing Cybe secu i y Th ough AI-Powe ed Solu ions: A Comp ehensi e Resea ch
Analysis. In e na ional Me idian Jou nal, 5(5), 1-8. h ps://me idianjou nal.in/index.php/IMJ/a icle/ iew/21
[10] Vegesna, V. V. (2023). Comp ehensi e analysis o AI-enhanced de ense sys ems in cybe space. In e na ional
Nume ic Jou nal o Machine Lea ning and Robo s, 7(7). h ps://injm .com/index.php/ ew ew /a icle/ iew/21
[11] Hussain, A. (2024). AI-Enhanced Cybe secu i y: S eamlining Da a Pipelines and Fo i ying Cloud In as uc u e
in a Digi al-Fi s E a.
[12] babu Nu halapa i, S. (2023). AI-enhanced de ec ion and mi iga ion o cybe secu i y h ea s in digi al banking.
Educ. Adm. Theo y P ac ., 29(1), 357-368.
[13] Pa ke , O. (2020). AI-Enhanced Da abase Managemen : S eng hening Cybe secu i y o In elligen Da a
P o ec ion.
[14] Fa hia, A. (1924). AI-Enhanced Cybe secu i y: Machine Lea ning o Anomaly De ec ion in Cloud Compu ing.
[15] Thapa, P., & A junan, T. (2024). AI-Enhanced Cybe secu i y: Machine Lea ning o Anomaly De ec ion in Cloud
Compu ing. Qua e ly Jou nal o Eme ging Technologies and Inno a ions, 9(1), 25-37.
h ps:// ec o al.o g/index.php/QJETI/a icle/ iew/64
[16] Islam, S. M., Sa ka , A., Khan, A. O. R., Islam, T., Paul, R., & Ba i, M. S. (2024). AI-D i en P edic i e Analy ics o
Enhancing Cybe secu i y in a Pos -Pandemic Wo ld: A Business S a egy App oach. In e na ional Jou nal o
Mul idisciplina y Resea ch.
[17] Uma , H., & Abbas, A. (2022). AI-Powe ed Th ea In elligence: Enhancing Cybe secu i y wi h P edic i e Analy ics
and Machine Lea ning.
[18] Nu alapa i, P. Enhancing Cybe secu i y wi h AI-Machine Lea ning Techniques o Anomaly De ec ion and
P e en ion. h ps://www.ije c .com/pape s/07-01/enhancing-cybe secu i y-wi h-ai-machine-lea ning-
echniques.pd
[19] McCall, A. (2024). AI and Cybe secu i y: De ec ing and Mi iga ing Cybe Th ea s.
[20] Mo i, J. (2023). AI-D i en Cybe Resilience in C i ical In as uc u e: Enhancing Th ea P edic ion, De ec ion, and
Reco e y. Jou nal o Compu ing and In o ma ion Technology, 3(1). h ps://uni e se-
publishe .com/index.php/jci /a icle/ iew/32
[21] Bibi, P. (2022). A i icial In elligence and Da abase Secu i y: Cu ing-Edge Techniques o Cybe Th ea
Mi iga ion.
[22] Jack, P., & Hu y, R. Ad anced Asse Secu i y Managemen : Le e aging AI and ML o Cybe Th ea De ec ion and
Mi iga ion.
[23] Ale izos, L., & Dekke , M. (2024). Towa ds an AI-Enhanced Cybe Th ea In elligence P ocessing Pipeline.
Elec onics 2024, 13, 2021. Machine Lea ning o Cybe secu i y, 202.
[24] Alessand o, R., & Giulia, B. (2024). AI-Enhanced Cybe secu i y P oac i e Measu es agains Ransomwa e and
Eme ging Th ea s. Inno a i e: In e na ional Mul i-disciplina y Jou nal o Applied Technology, 2(11), 77-92.
h p://ep in s.umsida.ac.id/14765/
[25] Kaul, D., & Khu ana, R. (2021). AI o De ec and Mi iga e Secu i y Vulne abili ies in APIs: Enc yp ion,
Au hen ica ion, and Anomaly De ec ion in En e p ise-Le el Dis ibu ed Sys ems. Eigenpub Re iew o Science and
Technology, 5(1), 34-62.
[26] Vance, T. R. Examina ion o Applica ions o A i icial In elligence in Cybe secu i y: S eng hening Na ional
De ense wi h AI.
[27] Chukwunweike, J. N., P aise, A., Osamuyi, O., Akinsuyi, S., & Akinsuyi, O. (2024). AI and Deep Cycle P edic ion:
Enhancing Cybe secu i y while Sa egua ding Da a P i acy and In o ma ion In eg i y. In e na ional Jou nal o
Resea ch Publica ion and Re iews, 5(8).
[28] Gha a , A., A shad, A., Abbas, S., & Tahi , M. (2024). A i icial In elligence in In o ma ion Technology: Enhancing
E iciency, Secu i y, and Inno a ion A Desc ip i e Re iew. Spec um o enginee ing sciences, 2(3), 289-309.
h p://www.sesjou nal.com/index.php/1/a icle/ iew/48
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1605
[29] Edwa d, A. (2020). Le e aging AI o S eng hen Cybe secu i y and Mi iga e Ransomwa e Th ea s in Heal hca e.
[30] Sanka am, M., Roopesh, M., Rase i, S., & Nisha , N. (2024). A COMPREHENSIVE REVIEW OF ARTIFICIAL
INTELLIGENCE APPLICATIONS IN ENHANCING CYBERSECURITY THREAT DETECTION AND RESPONSE
MECHANISMS. Managemen , 3(5).
[31] Malik, S. (2024). AI-Powe ed Cybe Risk Assessmen : P edic ing Vulne abili ies and A ack Vec o s in Real-Time.
[32] Ewan, P. E. (2024). Cybe secu i y F amewo k o Assessing he E iciency o AI-Based In usion De ec ion
Cybe secu i y Techniques (Doc o al disse a ion, Na ional Uni e si y).
h ps://sea ch.p oques .com/open iew/3906c565e7 6930a11919 9e9e5dde44/1?pq-
o igsi e=gschola &cbl=18750&diss=y
[33] 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. In e na ional Jou nal o In o ma ion
Technology and Elec ical Enginee ing (IJITEE)-UGC Ca e Lis G oup-I, 13(3), 11-19.
[34] Bhale ao, S., P abhu, S., & Ashok, P. (2024, Decembe ). AI Enabled Risk Managemen F amewo k o Enhanced
Secu i y in 5G Ne wo ks. In 2024 In e na ional Con e ence on Inno a ion and No el y in Enginee ing and
Technology (INNOVA) (Vol. 1, pp. 1-6). IEEE. h ps://ieeexplo e.ieee.o g/abs ac /documen /10847018/
[35] Oloyede, J. (2024). AI-D i en Cybe secu i y Solu ions: Enhancing De ense Mechanisms in he Digi al E a.
A ailable a SSRN 4976103. h ps://pape s.ss n.com/sol3/pape s.c m?abs ac _id=4976103
[36] Familoni, B. T. (2024). Cybe secu i y challenges in he age o AI: heo e ical app oaches and p ac ical solu ions.
Compu e Science & IT Resea ch Jou nal, 5(3), 703-724. h ps:// epbl.com/index.php/csi j/a icle/ iew/930
[37] Pa ke , O. (2020). AI and Cybe secu i y in Mode n Da abases: Inno a i e App oaches o Th ea De ec ion and
Response.
[38] Jabba , H., Al-Janabi, S., & Syms, F. (2024, Decembe ). AI-In eg a ed Cybe Secu i y Risk Managemen F amewo k
o IT P ojec s. In 2024 In e na ional Jo danian Cybe secu i y Con e ence (IJCC) (pp. 76-81). IEEE.
h ps://ieeexplo e.ieee.o g/abs ac /documen /10847294/
[39] Nicki, P. (2022). Real-Time Th ea In elligence: AI-Based App oaches o Cybe Risk P edic ion and Mi iga ion.
[40] Yaseen, A. (2023). AI-d i en h ea de ec ion and esponse: A pa adigm shi in cybe secu i y. In e na ional
Jou nal o In o ma ion and Cybe secu i y, 7(12), 25-43.
h ps://publica ions.dlp ess.o g/index.php/ijic/a icle/ iew/73
[41] Ali, H., & Zhang, S. (2020). AI-D i en Ne wo k Secu i y and Big Da a Analy ics: Imp o ing P oac i e De ense
S a egies in Cybe secu i y.
[42] Pe o ic, N., & Jo ano ic, A. (2023). Towa ds Resilien Cybe In as uc u e: Op imizing P o ec ion S a egies
wi h AI and Machine Lea ning in Cybe secu i y Pa adigms. In e na ional Jou nal o In o ma ion and
Cybe secu i y, 7(12), 44-60. h ps://publica ions.dlp ess.o g/index.php/ijic/a icle/ iew/75
[43] Luca, K., & Elena, F. (2023). THE INTERSECTION OF CYBERSECURITY AND AI LEVERAGING ARTIFICIAL
INTELLIGENCE TO MITIGATE EMERGING THREATS. Syne gy: C oss-Disciplina y Jou nal o Digi al In es iga ion,
1(1), 99-114. h p://ep in s.umsida.ac.id/14766/
[44] Todupunu i, A. (2023). The ole o a i icial in elligence in enhancing cybe secu i y measu es in online banking
using AI. In e na ional Jou nal o Enhanced Resea ch in Managemen & Compu e Applica ions, 12(01), 10-
55948. h ps://schola 9.com/publica ion/e26be2ba1c2420 1326becc52794d6db.pd
[45] Khan, R. S., Si azy, M. R. M., Das, R., & Rahman, S. (2022). An ai and ml-enabled amewo k o p oac i e isk
mi iga ion and esilience op imiza ion in global supply chains du ing na ional eme gencies. Sage Science Re iew
o Applied Machine Lea ning, 5(2), 127-144.
[46] Al u hood, B. S., Mankame, D. P., Dwi edi, M., & Jindal, M. N. A i icial In elligence and Cybe secu i y: Inno a ions,
Th ea s, and De ense S a egies.
[47] K einb ink, J. L. (2019). Analysis o a i icial in elligence (AI) enhanced echnologies in suppo o cybe de ense:
Ad an ages, challenges, and conside a ions o u u e deploymen (Mas e 's hesis, U ica College).
h ps://sea ch.p oques .com/open iew/2ca10115b5be484 c619b2534e01ace0/1?pq-
o igsi e=gschola &cbl=18750&diss=y
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1585-1606
1606
[48] Mazhe , N., Basha a , A., & Nisha , A. (2024). AI-D i en Th ea De ec ion: Re olu ionizing Cybe De ense
Mechanisms. Eas e n-Eu opean Jou nal o Enginee ing and Technology, 3(1), 70-82.
h p://snmzpublishe .com/index.php/EJET/a icle/ iew/127
[49] A junan, G. AI-Powe ed Cybe secu i y: De ec ing and P e en ing Mode n Th ea .
[50] Rockey, H. (2022). AI-D i en Cybe secu i y in IoT: De ec ing and P e en ing A acks on Sma De ices.
[51] Bibi, P. (2022). A i icial In elligence in Cybe secu i y: Re olu ionizing Da abase Managemen o Enhanced
P o ec ion.
[52] Hussain, S., & Elson, A. (2024). Ad e sa ial Machine Lea ning: Iden i ying and Mi iga ing AI-Powe ed Cybe
A acks.
[53] Raja, G., & Bai s ow, J. (2022). Ha nessing AI o Cybe secu i y: A Machine Lea ning App oach o Th ea De ec ion
and Da a P o ec ion.
