Co esponding au ho : Dinesh Deckke ORCID - 0009-0003-9968-5934
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.
A i icial In elligence and po nog aphy: A comp ehensi e esea ch e iew
Dinesh Deckke 1, * and Subhashini Sumanaseka a 2
1 MSc S uden , Depa men o Compu ing, W exham Uni e si y, Uni ed Kingdom.
2 MSc S uden , Depa men o Compu ing, Uni e si y o Glouces e shi e, Uni ed Kingdom.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
Publica ion his o y: Recei ed on 27 Ma ch 2025; e ised on 03 May 2025; accep ed on 06 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.2.1739
Abs ac
This comp ehensi e e iew examines he in e sec ion o a i icial in elligence (AI) and po nog aphy, analyzing how AI-
d i en echnologies such as deep akes, ecommenda ion sys ems, and con en mode a ion ools a e eshaping he adul
en e ainmen indus y. While AI in oduces e iciencies in con en c ea ion and pe sonaliza ion, i also gene a es
signi ican e hical, psychological, legal, and socie al challenges. The p oli e a ion o non-consensual deep ake
po nog aphy aises u gen conce ns abou consen , p i acy, and image-based sexual abuse. AI's ole in in luencing use
beha iou , ein o cing un ealis ic sexual no ms, and al e ing pe cep ions o in imacy is explo ed h ough psychological
and media e ec s heo ies. Addi ionally, he pape highligh s gaps in global egula ion, inconsis encies in legal
en o cemen , and he u gen need o longi udinal and in e en ion s udies o assess he eal-wo ld impac s o AI-
enhanced po nog aphy. Fu u e di ec ions emphasise he de elopmen o e hical amewo ks, obus echnological
sa egua ds, and in e disciplina y esea ch o guide esponsible inno a ion and p o ec human digni y in digi al
en i onmen s.
Keywo ds: A i icial In elligence; Po nog aphy; Deep akes; E hics; Consen ; P i acy; Recommenda ion Sys ems;
Con en Mode a ion; Socie al Impac
1. In oduc ion
1.1. O e iew o AI E olu ion and Applica ions
A i icial In elligence (AI) echnology has de eloped quickly in o an in luen ial ans o ma ional echnology a ec ing
heal hca e and inance sec o s, along wi h educa ion and digi al media. The capabili ies o AI s em om ad ancemen s
in machine lea ning (ML), along wi h deep lea ning and gene a i e modelling, which acili a e con en gene a ion, da a
analysis, and eal- ime in e ac ions, acco ding o Anan asi ichai & Bull (2021). The de elopmen o CNNS, GANS, and
di usion models allows de ices o gene a e au hen ic ex , images, and ideos wi h minimal human o e sigh (Gamage
e al., 2022; Blancha d & Taddeo, 2023). Technological ad ancemen s ha e gene a ed applica ions o c ea i e pu poses
and su eillance needs while disin eg a ion occu s be ween genuine and a i icial media (Velasco, 2022).
The essen ial unc ionali y o a i icial in elligence consis s o au oma ing labou -in ensi e wo k ac i i ies as well as
maximising decision suppo and con en il e ing capabili y (Gillespie, 2020). AI mode a ion sys ems cu en ly moni o
pla o ms o block explici con en alongside illegal o ha m ul pos s. Howe e , echnical issues a ound algo i hm bias
and inco ec con en unde s anding con inue o a ec he sys em's e ec i eness (Lai e al., 2022). AI de elopmen
ans o ms human in o ma ion and media in e ac ions along wi h human-human ela ions, gene a ing oppo uni ies
and aising e hical dilemmas.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
619
1.2. Po nog aphy as a Signi ican Segmen o Digi al Con en
Digi al media con ains po nog aphy as a signi ican segmen , which shapes cul u al a i udes and d i es consume
beha iou , oge he wi h pushing echnological adop ion. The adul indus y se ed as an ini ial d i e o mains eam
echnology accep ance since he popula i y o VHS and DVD up o he in oduc ion o online s eaming (Paasonen,
2011).
In e ne po nog aphy gene a es a dominan sha e o online a ic since adul con en websi es emain among he mos
isi ed websi es wo ldwide (Vincen , 2023). Inc eased in es iga ion in o p i acy iola ions, age au hen ici y assu ance,
and social consequences e alua ion has eme ged due o widesp ead adul con en use (Dasil a e al., 2021).
Digi isa ion has in ensi ied he accessibili y challenges while c ea ing e hical p oblems o he indus y. The
combina ion o anonymous online con en sha ing and he con enience o sha ing con en has esul ed in an inc ease
o non-consensual media along wi h e enge po n and AI-c ea ed in ima e images. E ol ing ends p esen challenges
o cu en egula o y amewo ks, necessi a ing p omp p o ec ion measu es o shield ulne able use s, as Wagne
and Blewe (2019) s a ed. The digi al po n en i onmen solidi ied i sel as he key ield o explo ing bo h media e hical
ma e s alongside use conduc and new echnology e ec s esea ch.
1.3. In e sec ion o AI and Po nog aphy
The con e gence o A i icial In elligence echnology wi h po nog aphy ep esen s a signi ican echnological as well as
e hical bounda y. AI sys ems domina e he adul indus y h ough hei capabili ies in making con en , deli e ing
pe sonalised ecommenda ions, and p o iding mode a ion ools. Gene a i e models employing Gene a i e Ad e sa ial
Ne wo ks (GANS) and di usion ne wo ks, in conjunc ion wi h cha bo s and i ual in luence s, acili a e he c ea ion o
high-de ini ion sexual media ia a i icial ex commands. This p ocess also esul s in he eme gence o con lic ing digi al
ela ionships (Hea en, 2023; Vincen , 2023). These echnological ad ancemen s acili a e high-dynamic con en
pe sonalisa ion and educe he need o human pe o me s, po en ially dec easing ce ain ypes o abuse.
AI-powe ed sexual con en c ea ion sys ems gene a e p o ound e hical ques ions ha p ima ily a ise om issues
in ol ing consen , along wi h p i acy conce ns and au hen ica ing iden i y. Deep akes, which desc ibe po nog aphic
con en c ea ed h ough AI body swapping p og ams wi hou consen , con inue o gain p ominence as a new o m o
abuse (Ci on, 2019; Ajde e al., 2019).
The la ge-scale p oduc ion o syn he ically made non-consensual con en h ough a i icial in elligence endange s
indi idual digni y while making legal esponsibili y di icul o de e mine (Sa a iano & Pe l o h, 2023). The ad ancing
po nog aphy echnology equi es an in e p o essional dialogue be ween legal sys ems and expe panels ocusing on
e hical p ac ices and men al well-being p o ec ion.
1.4. Objec i es o he Re iew Pape
1.4.1. To Syn hesise Exis ing Resea ch on AI Applica ions in Po nog aphy
This e iew aims o collec and examine exis ing schola ly ma e ial and legal documen s, as well as echnological s udies
ela ing o a i icial in elligence uses in po nog aphy. Resea che s ha e s a ed documen ing he ans o ma i e
capabili ies o a i icial in elligence echnologies ha now in ade all ace s o adul en e ainmen , acco ding o
Anan asi ichai & Bull (2021) and Hea en (2023). By syn hesising cu en s udies, his pape p o ides a comp ehensi e
unde s anding o how AI ools, such as gene a i e models, acial ecogni ion so wa e, na u al language p ocessing, and
algo i hmic ecommenda ion sys ems, a e used in a ious ace s o he adul indus y. I also seeks o d aw connec ions
be ween eme ging echnological capabili ies and hei e hical, legal, and psychological implica ions.
1.4.2. To Iden i y Gaps and Challenges in Cu en Knowledge
Despi e inc eased schola ly in e es in AI-gene a ed po nog aphy, he e emain c i ical gaps in bo h empi ical da a and
heo e ical amewo ks. Fo ins ance, esea ch on he long- e m psychological e ec s o hype -pe sonalised, AI-
gene a ed adul con en is s ill limi ed (Blancha d & Taddeo, 2023). Mo eo e , while s udies ha e documen ed he
sp ead and dange s o deep akes and syn he ic sexual media, pa icula ly hei non-consensual use, ew ha e
sys ema ically examined how hese ools impac ic ims’ men al heal h o al e public us in digi al image y (Sa a iano
& Pe l o h, 2023; Ci on, 2019). AI mode a ion ools and hei algo i hms expe ience nume ous echnical issues and
bias p oblems when p ocessing mino i y popula ions and ma ginal sexual con en , acco ding o Lai e al. (2022). The
en o cemen and p o ec ion sys ems unde global legal amewo ks show inconsis ency because hey all behind he
apid e olu ion o echnology (Velasco, 2022).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
620
1.4.3. To P opose Di ec ions o Fu u e Resea ch
The e iew guides academic in es iga ions and policy de elopmen by iden i ying esea ch di ec ions mo i a ed by
con empo a y e hical s anda ds. S udies ocusing on ex ended AI po nog aphy exposu e mus ollow indi iduals o
e ec analysis (Wagne & Blewe , 2019; Blancha d & Taddeo, 2023). The esea ch mus also in es iga e pla o m
mode a ion mechanisms and policies h ough expe imen al design (Go wa e al., 2020). Las ly, in e disciplina y e hical
models mus be es ablished o sa e AI use in adul con en gene a ion (Rui e , 2021). The analysis o AI-gene a ed
sexual con en egula ions ac oss mul iple ju isdic ions and he c ea ion o s anda dized s anda ds ha p o ec p i acy
and consen equi es u he comp ehensi e esea ch. Fu u e esea ch should ocus on esol ing hese p oblems o
c ea e an equilib ium be ween inno a i e de elopmen s and accoun abili y mechanisms in he de eloping ield o AI
po nog aphy.
1.5. Resea ch Impo ance
1.5.1. Unde s anding E hical, Psychological, and Socie al Implica ions
The in e sec ion o a i icial in elligence and po nog aphy p esen s a ious signi ican e hical issues, along wi h
psychological challenges and social dilemmas ha equi e u gen academic in es iga ion. The e hical p oduc ion and
dis ibu ion o AI-gene a ed po nog aphic ma e ials, including deep akes and syn he ic non-consensual media,
con adic s basic p inciples such as au onomy and consen as well as pe sonal digni y, acco ding o Rui e (2021) and
Ci on (2019). Deep ake po nog aphy ic ims deal wi h emo ional auma, oge he wi h epu a ional loss and digi al
iden i y con ol diminu ion, e en hough he images a e en i ely ake (Blancha d & Taddeo, 2023). When unau ho ised
con en depic ing mino s, public igu es, o p i a e pe sons appea s online, i uels s ong conce ns ha blu ee
exp ession limi s and digi al abuse guidelines (Sa a iano & Pe l o h, 2023).
The psychological exposu e o li elike AI-gene a ed po nog aphy has he po en ial o eshape use s' sexual
unde s anding; howe e , i dis o s hei pe cep ions o body shapes wi hin he a ious dimensions o human
ela ionships. Resea ch shows ha AI-enhanced adul con en wo sens an asy expec a ions by expanding objec i ying
con en he e o e p omo ing emo ional dis ance while leading people owa ds excessi e po n use (Wagne & Blewe ,
2019; Pawelec, 2022). The g ow h o AI-gene a ed e o ic cha bo s and i ual companions in luences human in imacy
de elopmen , which leads o ans o med socie al s anda ds abou in imacy and sexual p e e ence (Vincen , 2023).
AI-d i en po nog aphic con en no malisa ion will shape collec i e pe spec i es abou consen while changing public
iews ega ding au hen ici y, in addi ion o sexuali y. Deep akes c ea e such a usion o ic ion and eali y ha hey
diminish us in isual e idence and public o ums (Gamage e al., 2022). The wide cul u al e ec s o deep akes in ol e
al e ing e hical alues, and consume s become mo e exposed o exploi a ion, pa icula ly a ec ing ulne able
popula ion g oups who ace highe isks o exploi a i e media use (Lai e al., 2022).
1.5.2. In o ming Policy De elopmen and Technological Ad ancemen s
The esea ch guides egula o y agencies and echnological de ence mechanisms o mi iga e AI- ela ed isks associa ed
wi h adul con en . Va ious ju isdic ions possess di e en egula ions abou AI-gene a ed po nog aphy and deep akes
since p esen laws end o be agmen ed and delayed in hei esponses (Velasco, 2022). The Uni ed Kingdom and
Sou h Ko ea ha e es ablished laws p ohibi ing non-consensual AI con en sha ing, bu wo ldwide coope a ion on his
opic emains minimal (Kang, 2024; UK Home O ice, 2024). S udies abou AI po nog aphy e hics and legal aspec s will
help gene a e policies which e ec i ely p o ec igh s while enabling echnological de elopmen .
The ield's echnological indings can guide he de elopmen o AI sys ems owa d e hical s anda ds while concen a ing
on open sys em compliance, pe mission au hen ica ion, and con en il e ing. Fo ins ance, imp o ing AI de ec ion ools
o deep akes and implemen ing wa e ma king o aceabili y mechanisms could aid in dis inguishing au hen ic con en
om syn he ic ab ica ions (Blancha d & Taddeo, 2023). Fu he mo e, unde s anding he psychological impac o AI-
gene a ed po nog aphy can in o m he esponsible de elopmen o pla o ms and educa ional ini ia i es ha p omo e
digi al li e acy and espec ul online beha iou (Sau a e al., 2022).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
621
Figu e 1 F amewo k o AI in luence on Po nog aphic con en
This esea ch suppo s he sa e , mo e e hical in eg a ion o AI echnologies in o he digi al con en ecosys em by
iden i ying ha ms, p oposing solu ions, and encou aging in e disciplina y collabo a ion.
1.6. Resea ch Ques ions
• Wha a e he cu en applica ions o a i icial in elligence in he p oduc ion, pe sonalisa ion, and mode a ion
o po nog aphic con en ?
• Wha e hical and legal issues a e associa ed wi h he c ea ion and dis ibu ion o AI-gene a ed po nog aphy,
pa icula ly deep akes?
• How does AI-gene a ed po nog aphy impac psychological well-being and use beha iou o e ime?
• Wha a e he socie al implica ions o AI-enhanced adul con en , including e ec s on no ms, in imacy, and
consen ?
• How do AI-d i en ecommenda ion sys ems in luence use engagemen and aise p i acy conce ns in adul
en e ainmen pla o ms?
• How e ec i e a e cu en legal amewo ks and in e na ional egula ions in add essing he challenges posed
by AI-gene a ed explici con en ?
• Wha echnological and policy in e en ions ha e been p oposed o implemen ed o educe ha m associa ed
wi h AI-enhanced po nog aphy, and how e ec i e a e hey?
• Wha heo e ical amewo ks (e.g., echnological de e minism, media e ec s heo y, e hical models) bes
explain he in luence o AI in digi al sexual con en ?
• Wha a e he gaps in exis ing li e a u e, and how can u u e longi udinal and in e en ion s udies add ess
hem?
2. Me hodology
This s udy adop s a na a i e e iew app oach, s uc u ed by sys ema ic e iew p inciples, o explo e he in e sec ion
o a i icial in elligence (AI) and po nog aphy. While he e iew is na a i e in i s syn hesis, i inco po a es a
anspa en and eplicable p ocess o s udy selec ion and da a analysis o enhance eliabili y. The me hodology is
guided by he P e e ed Repo ing I ems o Sys ema ic Re iews and Me a-Analyses (PRISMA) amewo k, adap ed o
quali a i e syn hesis (Page e al., 2021).
2.1. Resea ch Design
The e iew ollowed an adap ed PRISMA p o ocol o ensu e me hodological anspa ency, minimize selec ion bias, and
main ain eplicabili y in iden i ying, sc eening, and including ele an li e a u e. Al hough no a ull me a-analysis, his
na a i e e iew main ains sys ema ic igou by o ganising he e iew p ocess a ound PRISMA's co e componen s:
eligibili y c i e ia, in o ma ion sou ces, sea ch s a egy, selec ion p ocess, and da a syn hesis.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
622
2.1.1. Sea ch S a egy
Iden i ica ion o Rele an Da abases
To cap u e he in e disciplina y na u e o he opic, he ollowing academic da abases we e selec ed o hei ex ensi e
co e age o bo h echnological and social science li e a u e: PubMed, IEEE Xplo e, Scopus, Google Schola and Web o
Science. These pla o ms p o ide access o pee - e iewed publica ions ac oss heal h, compu ing, e hics, psychology,
and communica ion s udies.
Sea ch Te ms and Boolean Ope a o s
A s uc u ed sea ch s ing was applied o each da abase using Boolean logic. The co e e ms we e:
("A i icial In elligence" OR "AI" OR "Machine Lea ning" OR "Deep Lea ning") AND ("Po nog aphy" OR "Adul Con en "
OR "Deep akes")
This o mula ion aimed o cap u e s udies discussing echnological de elopmen s (AI, ML, DL) wi hin digi al sexual
con en , including syn he ic media, mode a ion ools, and e hical/legal implica ions.
2.1.2. Inclusion and Exclusion C i e ia
The inclusion and exclusion c i e ia we e de ined o ensu e he ele ance and quali y o he li e a u e:
• Inclusion C i e ia: Pee - e iewed jou nal a icles, published in English, om 2010 onwa d, ocusing on AI
applica ions ela ed o po nog aphy, adul con en , o deep ake echnologies.
• Exclusion C i e ia: Non-pee - e iewed li e a u e (e.g., opinion pieces, news a icles), non-English publica ions,
and s udies un ela ed o he in e sec ion o AI and po nog aphy (e.g., gene al AI applica ions wi hou ele ance
o adul con en ).
These c i e ia ensu ed he e iew ocused on schola ly e idence add essing he subjec 's e hical, legal, echnological,
and psychological aspec s.
2.1.3. Da a Ex ac ion and Syn hesis
Key hemes, me hodologies, and indings we e sys ema ically ex ac ed om he included s udies. The da a ex ac ion
p ocess ocused on iden i ying:
• The ype o AI echnologies used (e.g., GANs, NLP, CNNs)
• Thei applica ion a eas (e.g., con en c ea ion, mode a ion, pe sonaliza ion)
• The scope o e hical, psychological, and legal implica ions
• Policy esponses and egula o y challenges
A quali a i e hema ic syn hesis was hen conduc ed o de ec ecu ing ends, heo e ical pe spec i es, and esea ch
gaps. This in ol ed i e a i e coding and compa ison ac oss s udies, aiming o p esen a comp ehensi e iew o he
cu en s a e o knowledge and he b oade social consequences o AI-gene a ed adul con en .
2.1.4. Limi a ions
Se e al limi a ions mus be acknowledged:
• Publica ion Bias: S udies wi h signi ican o no el indings a e mo e likely o be published, po en ially excluding
null o con adic o y e idence.
• Da abase Limi a ions: Al hough comp ehensi e, he selec ed da abases may no encompass all ele an wo k,
especially eme ging g ey li e a u e o egion-speci ic legal s udies.
• Subjec i e In e p e a ion: As a na a i e e iew, he syn hesis elies on he au ho s' in e p e a ion, which may
in oduce subjec i i y despi e e o s o ensu e balanced and c i ical analysis.
Add essing hese limi a ions in u u e esea ch could in ol e expanding da abase co e age, including mul ilingual
sou ces, o conduc ing me a-analy ical ollow-ups whe e possible.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
623
3. Li e a u e Re iew
3.1. His o ical Con ex
3.1.1. E olu ion o Po nog aphy in Digi al Media
Po nog aphy has unde gone a signi ican ans o ma ion wi h he ise o digi al media, e ol ing om p in and analogue
o ma s o online s eaming pla o ms, mobile apps, and in e ac i e con en deli e y sys ems. This ansi ion has no
only b oadened accessibili y bu also ede ined he modes o consump ion, enabling on-demand and pe sonalized adul
en e ainmen expe iences (Paasonen, 2011; Dasil a e al., 2021). The in e ne has played a pi o al ole in globalizing
po nog aphy, e oding p e ious geog aphical and egula o y ba ie s. Today, po nog aphic websi es consis en ly ank
among he mos isi ed on he in e ne , su passing many mains eam media pla o ms in a ic olume (Vincen , 2023).
The digi isa ion o po nog aphy has also c ea ed new e hical and legal challenges, pa icula ly conce ning use p i acy,
age e i ica ion, and unau ho ised con en dis ibu ion (Wagne & Blewe , 2019). The shee olume o con en and he
anonymi y o use s online ha e acili a ed he ise o e enge po n, ama eu uploads, and non-consensual sha ing. These
dynamics ha e posi ioned he adul en e ainmen indus y as a echnological inno a o and a ocal poin o digi al
go e nance deba es (Gillespie, 2020).
3.1.2. Ea ly Use o AI Technologies in Adul Con en
A i icial In elligence (AI) echnologies began in eg a ing in o he adul en e ainmen indus y in he ea ly 2010s,
p ima ily h ough basic ools o con en ca ego iza ion, acial ecogni ion, and ecommenda ion sys ems
(Anan asi ichai & Bull, 2021). These ea ly implemen a ions mi o ed de elopmen s in mains eam digi al pla o ms,
such as YouTube o Ne lix, whe e algo i hms enhanced use expe ience by p edic ing p e e ences and ailo ing con en
deli e y (King e al., 2019). Fo adul si es, such algo i hmic cu a ion helped s eamline na iga ion h ough as lib a ies
o con en , shaping use habi s and inc easing pla o m e en ion a es.
AI de eloped con en -making capabili ies s a ing om basic ace- ans o ming echnologies, which e en ually led o
he c ea ion o deep ake po nog aphy. Due o hei abili y o p oduce unsolici ed po nog aphy, FakeApp and DeepNude
spa ked signi ican con o e sy, while deep ake so wa e was ega ded as mo e expe imen al and niche (Ajde e al.,
2019). Syn he ic media p oduc ion using deep lea ning, speci ically GANS, has aken o e he discussion abou AI in
adul con en , acco ding o Gamage e al. (2022).
AI sys ems we e simul aneously de eloped o au oma ed con en mode a ion because pla o ms s a ed implemen ing
machine lea ning algo i hms o iden i y explici o ha m ul con en on hei pla o ms. AI ools expe ienced challenges
ega ding accu acy and biased pe o mance when used wi hou enough human inspec ion o hei esul s (Lai e al.,
2022; Go wa e al., 2020). The ini ial de elopmen s be ween AI and adul con en ideos es ablished a i s s ep o deep
in eg a ion be ween a i icial in elligence and he adul indus y, while p edic ing cu en inno a ions and
con o e sies wi hin his domain.
3.2. AI and Con en Gene a ion
3.2.1. Explo a ion o Deep ake Technology
Deep ake echnology de elopmen has esul ed in majo changes o c ea ing and ci cula ing explici digi al con en .
GANS enable deep ake echnology o p oduce ealis ic ake ideos and images ha combine one pe son's aces wi h
ano he 's bodies (Gamage e al., 2022; Ajde e al., 2019). In he adul con en sec o , his echnology has been widely
applied o gene a e syn he ic po nog aphy, o en wi hou he consen o he indi iduals depic ed.
Deep ake po nog aphy i s gained a en ion in 2017 h ough online o ums whe e use s began pos ing manipula ed
ideos o celeb i ies. Since hen, access o open-sou ce ools like FakeApp and DeepNude has democ a ised he c ea ion
o deep akes, making i possible o non-expe s o gene a e explici con en wi h minimal e o (Wagne & Blewe ,
2019). Mo e ecen ly, ad anced pla o ms now o e ex - o-image and ex - o- ideo deep ake gene a ion, enabling
use s o c ea e en i ely syn he ic adul ma e ial om simple p omp s (Hea en, 2023).
While deep akes can be used o c ea i e exp ession o an asy explo a ion, hei use in po nog aphy, especially non-
consensual and malicious applica ions, has become a cen al conce n in AI e hics and media egula ion (Rui e , 2021).
The sophis ica ion o hese ools poses a challenge o con en e i ica ion, wi h ake ideos o en being indis inguishable
om au hen ic oo age o he a e age iewe (Gamage e al., 2022).
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
624
Figu e 2 Types o AI Applica ions in Adul Con en
3.2.2. E hical Implica ions o AI-Gene a ed Explici Ma e ial
AI-gene a ed explici con en c ea ion gene a es mul iple e hical p oblems ela ed o consen while leading o
exploi a ion and causing ha m. The inc easing ecogni ion shows ha deep ake po nog aphy made wi hou consen
should be ea ed as a se e e ins ance o image-based sexual abuse because i in ades p i acy and denies physical
in eg i y (Ci on, 2019; Blancha d & Taddeo, 2023). AI-c ea ed explici media di e s om ypical adul ilms because i
p oduces ake sexual con en ha includes no ac ual pa icipan s o hei consen p o ocols.
The ei ica ion p ocess in ol ing human bodies ep esen s a undamen al mo al p oblem. The au ho Rui e (2021)
explains ha deep akes con e human beings in o objec s ha ge manipula ed and objec i ied o pleasu e, e en
hough hey emain unawa e o hei use. Due o his cycle, powe inequali ies become s onge while gende -based
iolence deepens, mainly when women a e mos o en a ge ed in deep ake po nog aphy, acco ding o Ajde e al.
(2019).
The c ea ion o AI po nog aphy has igge ed ea s ega ding he p oduc ion o simula ed images depic ing child sexual
abuse iola ions. The abili y o gene a e illegal o mo ally o ensi e digi al con en challenges cu en laws and e hical
guidelines because no eal child en pa icipa e in hese ma e ials (Sa a iano & Pe l o h, 2023). These illici media
ma e ials diminish public ai h in media o ganisa ions, besides po en ially c ea ing accep ance o de ian conduc and
desensi ising iewe s (Blancha d & Taddeo, 2023).
The de ende s o AI sys ems sugges human pe o me emo al om po nog aphic con en o p e en exploi a ion,
which c ea es e hical al e na i es o adi ional po nog aphy (Danahe , 2020). This po en ial ad an age canno
e ec i ely minimise he mo e signi ican p oblems o abuse and decep i e con en and social damage because o he
absence o clea di ec ion.
3.3. Pe sonalisa ion and Recommenda ion Sys ems
3.3.1. AI-D i en Recommenda ion Sys ems in Adul En e ainmen
Digi al adul en e ainmen pla o ms use AI-d i en ecommenda ion sys ems as hei main ea u e o p edic use
p e e ences, which hen deli e s cus omised con en expe iences. These sys ems' algo i hms use collabo a i e il e ing
and deep neu al ne wo ks o examine use da a consis ing o sea ch que ies and iewing ime wi h in e ac ion pa e ns
(Anan asi ichai & Bull, 2021). The algo i hms de ec pa e ns in use beha iou o deli e indi idually cus omised
con en , op imising cus ome engagemen while boos ing sha eholde e u ns (King e al., 2019).
The adul en e ainmen indus y deploys i s ecommenda ion sys ems wi h inc eased agg essi eness and sec ecy
because he e a e ew con en cons ain s and s ong consume demand o esh con en . Pla o ms apply AI
echnology o modi y ideo humbnails, me ada a, and sea ch esul algo i hms o achie e maximum use sa is ac ion
(Sau a e al., 2022). Acco ding o Pa ise (2011), he ex eme pe sonalisa ion p ocess in adul con en consump ion
c ea es il e bubbles h ough con inuous exposu e o inc easingly se e e sexual ma e ial.
These use expe ience op imisa ion sys ems gene a e unexpec ed ou comes, including he s eng hening o a e sexual
kinks and he no malisa ion o undesi able sexual ac s, while po en ially dis ega ding mino i y sexual o ien a ions
(Blancha d & Taddeo, 2023). The algo i hm's decision-making pa e ns p o e ha d o unde s and since use s s ay
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
625
unawa e o which pieces o con en he sys em displays o hem, hus spa king mul iple e hical inqui ies ega ding
pe sonal eedom and con ol.
3.3.2. Implica ions o Use Engagemen and P i acy Conce ns
The applica ion o a i icial in elligence echnologies in po nog aphy deli e s subs an ial inc eases in use dedica ion
and p olonged membe ship and con en iewing du a ion (Sau a e al., 2022). The com o p o ided by hese se ices
leads o p i acy isks o use s. The ope a ion o hese sys ems depends on la ge amoun s o sensi i e pe sonal da a,
including indi idual p e e ences, demog aphic in o ma ion, and beha iou al s a is ics. Ninja and eamma e analysis o
his da a esul s in p i acy b eaches, p o iling p ac ices and imp ope use o in o ma ion because companies ha e
insu icien sa egua ds (King e al., 2019).
The p ac ice o da a handling on adul pla o ms emains unclea o use s. The disclosu e o use da a and p ocessing
me hods emains unclea o mos cus ome s abou whe e hei in o ma ion is s o ed and dis ibu ed. Use p o ec ion
becomes insu icien because mos pla o ms ail o p o ide clea in o ma ion abou hei da a p ac ices, which yields
a eas o ulne abili y o b eaches, ad e ising dis up ion, and unau ho ised sha ing o use in o ma ion (Pawelec,
2022). Regions wi h insu icien da a p o ec ion laws encoun e mo e se e e issues due o inadequa e en o cemen o
legisla ion compa able o he Gene al Da a P o ec ion Regula ion (GDPR).
Resea che s ha e no ho oughly s udied he psychological e ec s o he cons an low o algo i hm-gene a ed con en .
Resea ch poin s owa ds pe sis en exposu e o pe sonalised adul con en as a cause o p oblema ic use beha iou
simila o addic ion, alongside a shi in pe cep ions owa d sexual expe iences and eal in e ac ions, and he
de elopmen o dis o ed iews abou sexuali y and pe sonal iden i y (Wagne & Blewe , 2019).
AI ecommenda ion sys ems equi e buil -in e hical p inciples, such as da a minimisa ion and anspa ency, consen
mechanisms, and op -ou op ions, o balance pe sonalisa ion and use p o ec ion.
3.4. AI in Con en Mode a ion
3.4.1. U ilisa ion o AI o Fil e ing Explici o Illegal Con en
Digi al pla o ms use A i icial In elligence (AI) as hei p ima y con en mode a ion sys em o il e all explici , non-
consensual, and illegal ma e ial. The con inuing apid expansion o use -gene a ed con en , pa icula ly in adul
en e ainmen , calls o mo e me hods han manual mode a ion o su ice. Emba king on ackling hei con en issues,
pla o ms employed a i icial in elligence classi ie s, compu e ision algo i hms, and na u al language p ocessing
sys ems o iden i y and emo e illegal o b each-o -con ac ma e ials, acco ding o Gillespie (2020) and Lai e al.
(2022).
AI sys ems o con en mode a ion p ocess la ge, ained da ase s o labelled da a o es ablish de ec ion capabili ies o
nudi y while iden i ying sexual ac i i y and iolen isual con en . These in o ma ion echnology sys ems enable
pla o ms o main ain compliance wi h po nog aphy- ela ed egula ions, which include age e i ica ions as well as
consen p o ocols and deep ake and child sexual abuse ma e ial (CSAM) es ic ions (Velasco, 2022; Go wa e al., 2020).
Compu e ision and NLP wo k oge he o pe o m image pa e n ecogni ion o explici ma e ial iden i ica ion and
me ada a and use commen e alua ion o coe ci e con ex ual de ec ion.
The ools o e pla o ms e icien moni o ing capabili ies o managing la ge olumes o objec ionable ma e ial ac oss
hei ne wo ks. The au ogene a ion o con en mode a ion p esen s challenges ega ding isible p ocesses, balanced
app oaches, and excessi e con ol when human in e en ion is una ailable o decision-making (Blancha d & Taddeo,
2023).
3.4.2. Challenges and Biases in Mode a ion Technologies
AI mode a ion echnologies encoun e a ious echnical limi a ions and e hical conce ns, e en hough hey ha e
become popula h oughou all indus ies. Algo i hmic bias emains one o he main p oblems because i a ises om
aining algo i hms wi h unbalanced da ase s ha e lec p e alen cul u al, acial, and gende no ms (Go wa e al.,
2020). AI censo ship echnology applies biased il e ing ha esul s in unequal ea men o mino i y con en c ea o s,
sexual wo ke s, and cul u al mani es a ions o sexuali y ou side Wes e n no ms (Sla e y e al., 2024).
AI sys ems gene a e alse posi i es h ough hei inabili y o di e en ia e be ween educa ional sexual heal h con en
and adul ma e ial, which leads o inapp op ia e con en emo als (Gillespie, 2020). The so wa e de ec ion sys em
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
626
c ea es alse nega i es by neglec ing o ind dange ous o exploi a i e con en , speci ically in cases whe e o ende s
modi y hei con en o e ade de ec ion.
The majo p oblem s ems om AI mode a ion ools, which ope a e as black boxes due o hei complexi ies. Use s do
no unde s and mode a ion sys ems' decision-making me hods and how o pa icipa e in mode a ion challenges. Use s'
con idence and esponsibili y diminish because o inadequa e anspa ency measu es in machine lea ning sys ems,
pa icula ly in si ua ions ha ha m a use 's epu a ion o income (Lai e al., 2022). Acco ding o Blancha d and Taddeo
(2023), AI echnology equi es e hical s anda ds ha mus emphasize ai ness, explainabili y, and he igh o seek
ed ess.
The capabili ies o AI ex end o managing high wo kloads and as p ocessing, bu comple e supe ision equi es human
in e en ion. Acco ding o Lai e al. (2022), he ecommended app oach o mode a ion combines human judgmen
wi h machine lea ning echnology in o hyb id models. The hyb id models iden i y e hical si ua ions and p o ide
con ex ual unde s anding, which helps p e en he ad e se impac s o comple e au oma ion sys ems.
3.5. AI and Use P i acy
3.5.1. P i acy Risks Associa ed wi h AI-Powe ed Adul Con en Pla o ms
The pe o mance and ecommenda ions o AI-powe ed adul con en pla o ms depend hea ily on collec ing use da a,
which enables pe sonalised expe iences and sys em op imiza ion. These p ocessing me hods c ea e majo p i acy
conce ns since he con en and beha iou da a ep esen highly con iden ial in o ma ion. Use s shed massi e quan i ies
o da a abou hei sea ches and iewing habi s, de ice in o ma ion, and IP numbe s o pla o ms e en hough hey do
no g asp he comple e ex en o da a acking p ocedu es (King e al., 2019; Sau a e al., 2022).
Th ough p edic i e analy ics and deep lea ning models, pla o ms gain he abili y o iden i y in ima e p e e ences and
beha iou al pa e ns, which leads o de ailed p o ile gene a ion ha pla o ms sell o exploi . The exposu e o adul
con en da a combined wi h unau ho ised su eillance o iden i y he e en s p esen s signi ican conce ns because
such b eaches migh esul in epu a ional ha m, blackmail inciden s, o psychological damage (Pawelec, 2022).
Collec ing inancial da a becomes mo e dange ous because i combines wi h sensi i e pe sonal in o ma ion in p emium
o subsc ip ion-based pla o ms.
Many adul websi es do no implemen p ac ical e hical s anda ds o da a p o ec ion ules, lea ing hei use s a isk o
unde ec able au oma ed p o iling and algo i hmic scheme con ol. The imp ope design o AI sys ems combined wi h
insu icien egula ion allows hese sys ems o b eak p i acy ules by de ec ing p i a e sexual p e e ences o use s and
linking di e en digi al iden i ies h ough acial ecogni ion o c oss-pla o m acking (Blancha d & Taddeo, 2023).
3.5.2. Da a P o ec ion Conce ns and Regula o y Responses
Inc easing conce ns abou p i acy ha e p omp ed he GDPR, he da a p o ec ion law in he Eu opean Union, and he
CCPA, he da a p o ec ion law in he Uni ed S a es, o c ea e sys ems ha allow use s be e con ol o e hei pe sonal
in o ma ion. The egula ions manda e pla o m ope a o s o ge di ec pe mission om use s abou da a p ac ices while
p o iding in o ma ion disclosu e abou hese p ocedu es and p o iding access o da a dele ion op ions (Go wa e al.,
2020).
Implemen ing new p i acy egula ions aces weak en o cemen s anda ds since pla o ms ope a ing ac oss bo de s
unc ion wi hin un egula ed digi al pla o ms. Adul con en p o ide s exploi weak da a p o ec ion laws o ju isdic ions
whe e hey ope a e o s ay unaccoun able and collec use in o ma ion in ex ensi e quan i ies (Velasco, 2022).
Alongside AI echnology, use s lose p o ec ion unde law because i s au oma ed da a ex ac ion p ocedu es and
decision-making algo i hms un unde ec ed. A i icial In elligence echnology analyses biome ic da a and emo ional
esponses ob ained om came as o moni o use s, he eby aising subs an ial conce ns ega ding he bounda ies o
use consen and he limi a ions o su eillance, as a icula ed by Sau a e al. (2022). Cu en egula o y measu es
emain inadequa e o p o ec agains de eloping dange s in adul con en because AI echnology deploymen s ou pace
exis ing egula o y s uc u es.
Digi al igh s g oups and esea che s suppo he need o unique p o ec ion guidelines speci ic o AI and p opose ules
ha su pass s anda d p i acy equi emen s. T acking e hical pe o mance in AI sys ems would inco po a e p o isions
o algo i hm isibili y and use con ol sys ems and ac i e da a sys em checks o e i y compliance (Blancha d &
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
633
and consen (B yan & Zillmann, 2002; Sho e al., 2012). E hical amewo ks—including u ili a ian, deon ological, and
i ue e hics models—p o ide no ma i e guidance o minimizing ha m, p o ec ing digni y, and ensu ing esponsible
AI deploymen (Rui e , 2021; Sau a e al., 2022). These heo e ical lenses unde sco e he u gen need o e hical
inno a ion in AI-d i en media.
4.9. Gaps in Exis ing Li e a u e and Fu u e Resea ch Needs (RQ9)
Signi ican gaps pe sis in he schola ly unde s anding o AI-gene a ed po nog aphy. Longi udinal esea ch assessing
syn he ic adul con en 's sus ained psychological and socie al e ec s emains sca ce (Sho e al., 2012; Vincen , 2023).
Despi e he global each o AI echnologies, c oss-cul u al compa isons o legal and e hical esponses a e also limi ed
(Velasco, 2022). Fu he mo e, ew in e en ion s udies e alua e he eal-wo ld e ec i eness o echnological o policy
sa egua ds. Fu u e esea ch should adop in e disciplina y, longi udinal, and pa icipa o y app oaches o unde s and
be e he beha iou al, legal, and e hical implica ions o AI-enhanced po nog aphy. Add essing hese gaps is c ucial o
de eloping comp ehensi e amewo ks ha balance echnological inno a ion wi h p o ec ing indi idual igh s and
socie al well-being.
Figu e 7 Cu en Knowledge Gaps in AI-Gene a ed Po nog aphy Resea ch
5. Fu u e Di ec ions
5.1. Longi udinal S udies
Resea ch abou he long- e m in luence o a i icial in elligence (AI) on socie y and psychology needs u u e s udy
because AI is s ill de eloping as a gen e in adul media. Resea ch li e a u e cu en ly exhibi s mos ly combined c oss-
sec ional s udies wi h me e heo ies abou how long- e m exposu e o AI-gene a ed adul con en al e s use s
h oughou speci ic du a ions (Sho e al., 2012).
Resea ch mus use longi udinal app oaches o ollow how exposu e o AI-gene a ed po nog aphy impac s sexual
sa is ac ion le els, along wi h sel -es eem measu es, in addi ion o compulsi e beha iou pa e ns and emo ional
de achmen . Resea ch mus expand beyond cu en ime ames o p o e he e ec s o hype -pe sonalised o syn he ic
con en exposu e on desensi isa ion, eal-li e in imacy pe cep ion, and maladap i e beha iou de elopmen (Vincen ,
2023; Pawelec, 2022).
Longi udinal esea ch se es a c i ical pu pose o s udy how changing e hical bounda ies and sexual no ms un old
ega ding AI-gene a ed po n h oughou socie y. These in es iga ions need o analyse how di e en gene a ions iew
sexual consen alongside he social changes o syn he ic sexual beha io s and digi al en i onmen p i acy s anda ds
(Blancha d & Taddeo, 2023; Sla e y e al., 2024).
This esea ch would assis policy de elopmen h ough e idence ha iden i ies popula ions mos ulne able o ha m,
such as adolescen s and people wi h sexual compulsions, and socially isola ed g oups, and p oposes echnological and
educa ional solu ions o p o ec ion. S udying he ans o ming ole o AI in human sexuali y demands eamwo k
be ween psychologis s and sociologis s, oge he wi h e hical expe s and echnologis s.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
634
Longi udinal esea ch allows us o o e come heo e ical p edic ions by acking he ac ual de eloping ou comes o AI-
based po nog aphy usage. Such amewo ks build s ong bases o di ec he making o e hical pla o ms and bo h he
egula ion o pla o ms and public awa eness e o s.
5.2. In e en ion S udies
AI p og ess in po nog aphic con en c ea ion, dissemina ion, and consump ion equi es immedia e esea ch on bo h
echnological solu ions and policy s a egies which will mi iga e po en ial ha m om AI-gene a ed po nog aphic
ma e ials. The exis ing ha ms su ounding deep akes s em om unwan ed dis ibu ion o non-consensual con en along
wi h p i acy iola ions and ex ensi e po nog aphy consump ion and e hical deg ada ion o audience a i udes
(Blancha d & Taddeo, 2023; Ci on, 2019).
Resea ch ini ia i es mus ocus on p oducing deep ake de ec ion sys ems alongside digi al wa e ma king me hods,
syn he ic media agging, and consen e i ica ion sys ems acco ding o echnological needs. Resea ch indica es ha
compu e lea ning sys ems ha can de ec AI-manipula ed con en h ough acial inconsis encies o me ada a
i egula i ies success ully iden i y manipula ed images (Gamage e al., 2022). These de ec ion ools need con inuous
imp o emen o ope a e e icien ly agains p og essi ely mo e p o essional a i icial in elligence models. Resea ch
in e en ions a e necessa y o e alua e he e ec i eness, p ecision, and eadiness- o-use me ics o hese echnologies
on li e pla o ms (Ajde e al., 2019).
Sys ema ic s udies mus assess policy esponses o hei ou comes. The Online Sa e y Ac o he Uni ed Kingdom,
alongside Sou h Ko ean and Eu opean Union legisla i e mo emen s, adop ed c iminalisa ion ules o non-consensual
deep akes (Kang, 2024; UK Home O ice, 2024). Resea ch s udies will e alua e he success o new policies h ough hei
powe o educe ha m while boos ing he accoun abili y o w ongdoe s and enhancing public awa eness. The
esea che s need o assess whe he cu en punishmen s a egies p o e adequa e o i supplemen al p og ams,
including public awa eness ini ia i es, ic im suppo p og ams, and con en mode a ion p o ocols, should be added.
The esea ch ocuses on de eloping s a egies o pla o m managemen . Con en -hos ing pla o ms se e as c ucial
en i ies which bo h deploy AI mode a ion sys ems and implemen e hical design p inciples, as well as ensu e
anspa en op ions o con en appeals. S udies abou in e en ion enable he disco e y o op imal p ac ices ha
main ain ee exp ession while p o ec ing agains ha m ul con en , pa icula ly o unusual cases, including pa odies,
sa i e, and con en p oduced by AI (Go wa e al., 2020; Lai e al., 2022).
Resea ch p ojec s ha uni e echnology p o essionals wi h legal expe s and e hical and psychological esea che s a e
i al o p oducing solu ions ha p o ide wide- eaching bene i s and lexibili y. S akeholde engagemen mus be he
co e elemen o hese s udies, which should in ol e deep ake abuse su i o s, de elope s, egula o s, and ci il socie y
ac o s.
Resea ch in e en ions need o become he cen al ocus because hey will help mo e cu en eac i e policies owa ds
p oac i e ha m educ ion measu es wi hin he age o AI-enhanced po nog aphy. The implemen ed app oaches will
assis in building sa e digi al spaces wi h s onge e hical s anda ds.
5.3. E hical F amewo ks
P o ec ing adul con en h ough obus e hical amewo ks is i al when implemen ing a i icial in elligence sys ems
in hei c ea ion, dis ibu ion, and use. The ield o a i icial in elligence po nog aphy now challenges basic s anda ds
o consen and p i acy alongside au hen ici y which equi es upda ed e hical ules o managemen .
The p inciple o in o med consen s ands as an essen ial e hical base in all guidelines. AI sys ems mus inco po a e
e hical ea u es ha equi e p oo o explici consen om indi iduals whose physical o ms o iden i y da a become
pa o digi al c ea ions (Bo h Ci on, 2019 and Rui e , 2021 ag ee). E hical sys ems mus p o ide people wi h dynamic
consen mechanisms ha gi e hem comple e con ol o cancel au ho isa ion o syn he ic con en gene a ion and
subsequen emo al. AI ools mus es ablish e ms o consen because hei absence would enable sexual abuse h ough
images and se e e a acks agains human digni y (Blancha d & Taddeo, 2023).
T anspa ency oge he wi h accoun abili y ep esen essen ial building blocks ha should be implemen ed. Con en
c ea o s along wi h pla o m ope a o s oge he wi h so wa e de elope s mus e eal ins ances whe e con en o igins
om AI sys ems o encoun e s modi ica ions (Ajde e al., 2019; Gamage e al., 2022) h ough isible ags o unique
digi al ma ke s o use s o ecognize a i icial con en om genuine eco dings. AI sys ems equi e buil -in
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
635
explainabili y ea u es so a ec ed pe sons along wi h egula o y bodies unde s and bo h he ope a ion o algo i hms
and hei impac on con en - ela ed decisions (Sau a e al., 2022).
Syn he ic digni y eme ges as an impo an e hical p inciple unde AI e hics policies o g an ing mo al conside a ion o
a i icially gene a ed depic ions ha show iden i iable people (Sla e y e al., 2024). All e hical sys ems should ecognise
and emedy he emo ional damage, epu a ional damage, and psychological auma om unau ho ised syn he ic
po ayals, e en i physical mee ings did no ake place (Blancha d & Taddeo, 2023).
The mo al philosophy o i uous conduc unc ions as a guide by ocusing on he mo al cha ac e o he de elope and
pla o m owne and hei du y o esponsibili y. AI sys ems a e de eloped o in eg a e e hical p o ec ion ea u es,
including bias de ec ion capabili ies, consen e i ica ion sys ems, and op -ou op ions, he eby demons a ing
dedica ion o indi idual au onomy and jus ice alongside empa hy (Danahe 2020; Rui e 2021).
E hical amewo ks equi e inclusi e and in e sec ional app oaches which p e en sys ema ic a ge ing o women,
LGBTQ+ indi iduals, acial mino i ies, as well as o he s who ace disp opo iona e non-consensual deep ake and
algo i hmic bias exposu e (Lai e al., 2022; Sla e y e al., 2024). The e hical s anda ds o p o ec ing ulne able
communi ies need o be de eloped h ough collec i e wo k be ween ic ims o digi al abuse and g oups such as
e hicis s, echnologis s, and human igh s ad oca es.
Building solid e hical amewo ks o adul con en A i icial In elligence equi es mo e han echnical ac ics because
i ep esen s an essen ial mo al du y. A p oac i e app oach combining mul iple disciplines unde p inciples o consen ,
anspa ency, digni y, accoun abili y and social jus ice p o ec s indi idual igh s while p o ec ing hei digi al well-
being in syn he ic digi al domains.
6. Conclusion
6.1. Summa y o Key Findings
This e iew syn hesises he cu en li e a u e on he in e sec ion o a i icial in elligence (AI) and po nog aphy,
e ealing bo h g oundb eaking applica ions and u gen e hical, psychological, and legal conce ns. AI is e olu ionising
adul con en p oduc ion, dis ibu ion, and consump ion h ough deep akes, ecommenda ion sys ems, and mode a ion
ools. While hese inno a ions inc ease e iciency and engagemen , hey aise signi ican issues a ound consen , iden i y,
p i acy, and long- e m psychological impac . The li e a u e also highligh s egula o y agmen a ion, echnological
opaci y, and longi udinal and in e en ion esea ch gaps. Below is a summa y o he key indings:
6.1.1. Summa y o Key Findings and Li e a u e Gaps
Resea ch Ques ion
Key Findings
Li e a u e Gaps
RQ1: Cu en applica ions o AI
in adul con en
AI enables syn he ic con en gene a ion
(deep akes, GANs), hype -pe sonalized
ecommenda ions, and con en mode a ion.
Limi ed empi ical s udies on eal-
wo ld e ec i eness and use
impac s.
RQ2: E hical and legal issues o
AI-gene a ed po nog aphy
Non-consensual deep akes iola e digni y and
p i acy; legal en o cemen is agmen ed
globally.
Lack o global s anda ds and
challenges add essing syn he ic
non-iden i iable con en .
RQ3: Psychological impac s
o e ime
AI con en os e s compulsi e use, un ealis ic
sexual expec a ions, and emo ional
de achmen .
The e is a need o longi udinal
s udies assessing long- e m
psychological impac s.
RQ4: Socie al implica ions on
no ms, in imacy, and consen
Syn he ic con en eshapes no ms o consen ,
au hen ici y, and human in imacy.
Minimal c oss-cul u al esea ch on
socie al shi s.
RQ5: In luence o AI-d i en
ecommenda ion sys ems and
p i acy isks
AI ecommenda ion sys ems enhance
engagemen bu comp omise use p i acy and
da a secu i y.
The e is a need o s onge p i acy
go e nance and anspa ency
esea ch.
RQ6: E ec i eness o cu en
legal amewo ks
Some coun ies c iminalise non-consensual AI
po n, bu global en o cemen emains
inconsis en .
Ju isdic ional loopholes and weak
c oss-bo de en o cemen
mechanisms.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
636
RQ7: Technological and policy
in e en ions
Deep ake de ec ion ools, wa e ma king, and
consen e i ica ion a e eme ging.
Few in e en ion s udies e alua e
eal-wo ld e ec i eness.
RQ8: Theo e ical amewo ks
explaining AI's in luence
Technological de e minism and media e ec s
heo y explain beha iou al changes.
Need o deepe in e disciplina y
in eg a ion o heo ies.
RQ9: Li e a u e gaps and u u e
esea ch
Subs an ial gaps in empi ical, longi udinal, and
c oss-cul u al esea ch.
Necessi y o pa icipa o y,
in e disciplina y s udies.
6.2. Call o Ac ion
U gen collabo a ion among esea che s, echnologis s, lawmake s, and ci il socie y membe s becomes essen ial
because o he deep consequences AI b ings o po nog aphy. This domain equi es in e disciplina y coope a ion
be ween p o essionals who s udy compu e science, law, psychology, e hics and media s udies, and sociology. The
esea ch communi y needs o conduc ex ensi e longi udinal s udies and in e ene ac oss beha io al domains o AI
po n phenomena in o de o p oduce e idence-based policy ecommenda ions.
A p oac i e policy amewo k mus be es ablished because eac i e en o cemen app oaches p o e insu icien when
dealing wi h he ma e . The es ablished amewo ks mus add ess syn he ic iden i y misuse while se ing consen
equi emen s and making con en pla o ms disclose hei ope a ions o use s. New echnological de ec ion solu ions
equi e de elopmen wi h undamen al human igh s p inciples and digni y cen al o hei s uc u e.
The apid de elopmen o AI echnologies wi hin adul con en equi es equally as de elopmen o esponsible
go e nance, e hical p o ocols, and public awa eness e o s. These e o s will de e mine he p o ec ion o digi al
in imacy beha iou and use consen .
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] Adle , A. (2024). A ousal by algo i hm: Sex ech and he ise o a i icial in imacy. Co nell Law Re iew, 109(4),
787–841.
[2] Ajde , H., Pa ini, G., Ca alli, F., & Cullen, L. (2019). The s a e o deep akes: Landscape, h ea s, and impac .
Deep ace. h ps://sensi y.ai/ epo s/
[3] Anan asi ichai, N., & Bull, D. (2021). A i icial in elligence in he c ea i e indus ies: A e iew. A i icial
In elligence Re iew, 54, 3311–3339. h ps://doi.o g/10.1007/s10462-021-10039-7
[4] Blancha d, A., & Taddeo, M. (2023). The e hics o a i icial in elligence o in elligence analysis: A e iew o he
key challenges wi h ecommenda ions. Digi al Socie y, 2(12). h ps://doi.o g/10.1007/s44206-023-00036-4
[5] B yan , J., & Zillmann, D. (2002). Media e ec s: Ad ances in heo y and esea ch (2nd ed.). Rou ledge.
[6] Chandle , D. (2002). Technological o media de e minism. h p:// isual-
memo y.co.uk/daniel/Documen s/ ecde / ecde .h ml
[7] Ci on, D. K. (2019). Sexual p i acy. Yale Law Jou nal, 128(7), 1870–1960.
[8] Danahe , J. (2017). The philosophical case o obo sex. In J. Danahe & N. McA hu (Eds.), Robo sex: Social and
e hical implica ions (pp. 3–28). MIT P ess.
[9] Dasil a, J. P., Aye di, K. M., & Galdoz, T. M. (2021). Deep akes on Twi e : Which ac o s con ol hei sp ead? Media
and Communica ion, 9(1), 170–181. h ps://doi.o g/10.17645/mac. 9i1.3433
[10] Gamage, D., Ghasiya, P., Bonagi i, V., Whi ing, M. E., & Sasaha a, K. (2022). A e deep akes conce ning? Analyzing
con e sa ions o deep akes on Reddi and explo ing socie al implica ions. P oceedings o he ACM Web
Con e ence 2022, 1001–1012. h ps://doi.o g/10.1145/3491102.3517446
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 618-637
637
[11] Gillespie, T. (2020). Con en mode a ion, AI, and he ques ion o scale. Big Da a & Socie y, 7(2).
h ps://doi.o g/10.1177/2053951720943234
[12] Go wa, R., Binns, R., & Ka zenbach, C. (2020). Algo i hmic con en mode a ion: Technical and poli ical challenges
in he au oma ion o pla o m go e nance. Big Da a & Socie y, 7(1).
h ps://doi.o g/10.1177/2053951719897945
[13] Hea en, W. D. (2023, Augus 1). These new ools le you gene a e po n om sc a ch. MIT Technology Re iew.
h ps://www. echnology e iew.com
[14] Kang, J. (2024, Feb ua y 10). Sou h Ko ea bans iewing o AI-gene a ed po n o eal people. The New Yo k Times.
h ps://www.ny imes.com
[15] King, T. C., Agga wal, N., Taddeo, M., & Flo idi, L. (2019). A i icial in elligence c ime: An in e disciplina y analysis
o o eseeable h ea s and solu ions. Science and Enginee ing E hics, 25(2), 489–509.
h ps://doi.o g/10.1007/s11948-018-00081-0
[16] Lai, V., Ca on, S., Bha naga , R., Liao, Q. V., Zhang, Y., & Tan, C. (2022). Human-AI collabo a ion ia condi ional
delega ion: A case s udy o con en mode a ion. P oceedings o he 2022 ACM CHI Con e ence on Human Fac o s
in Compu ing Sys ems. h ps://doi.o g/10.1145/3491102.3501999
[17] Paasonen, S. (2011). Ca nal esonance: A ec and online po nog aphy. MIT P ess.
[18] Page, M. J., McKenzie, J. E., Bossuy , P. M., Bou on, I., Ho mann, T. C., Mul ow, C. D., ... & Mohe , D. (2021). The
PRISMA 2020 s a emen : An upda ed guideline o epo ing sys ema ic e iews. BMJ, 372, n71.
h ps://doi.o g/10.1136/bmj.n71
[19] Pa ise , E. (2011). The il e bubble: Wha he In e ne is hiding om you. Penguin P ess.
[20] Pawelec, M. (2022). Deep akes and democ acy ( heo y): How syn he ic audio isual media o disin o ma ion and
ha e speech h ea en co e democ a ic unc ions. AI & Socie y. h ps://doi.o g/10.1007/s44206-022-00010-6
[21] Rui e , A. D. (2021). The dis inc w ong o deep akes. Philosophy & Technology, 34, 165–183.
h ps://doi.o g/10.1007/s13347-021-00459-2
[22] Sa a iano, A., & Pe l o h, N. (2023, Ma ch 17). A.I. images o child sexual abuse es limi s o he law. The New
Yo k Times. h ps://www.ny imes.com
[23] Sau a, J. R., So iano, D. R., & Palacios-Ma qués, D. (2022). Assessing beha io al da a science p i acy issues in
go e nmen a i icial in elligence deploymen . Go e nmen In o ma ion Qua e ly, 39(4), 101679.
h ps://doi.o g/10.1016/j.giq.2022.101679
[24] Sho , M. B., Black, L., Smi h, A. H., We e neck, C. T., & Wells, D. E. (2012). A e iew o in e ne po nog aphy use
esea ch: Me hodology and con en om he pas 10 yea s. Cybe psychology, Beha io , and Social Ne wo king,
15(1), 13–23. h ps://doi.o g/10.1089/cybe .2010.0477
[25] Sla e y, P., Sae i, A. K., G undy, E. A. C., G aham, J., Noe el, M., Uuk, R., Dao, J., Pou , S., Caspe , S., & Thompson, N.
(2024). The AI isk eposi o y: A comp ehensi e me a- e iew, da abase, and axonomy o isks om a i icial
in elligence. a Xi p ep in . h ps://doi.o g/10.48550/a Xi .2408.12622
[26] UK Home O ice. (2024). Online Sa e y Ac : P o ec ing people om ha m ul con en .
h ps://www.go .uk/go e nmen /collec ions/online-sa e y-bill
[27] Velasco, C. (2022). Cybe c ime and a i icial in elligence: An o e iew o he wo k o in e na ional o ganiza ions
on c iminal jus ice and he in e na ional applicable ins umen s. ERA Fo um, 23, 127–145.
h ps://doi.o g/10.1007/s12027-022-00702-z
[28] Vincen , J. (2023, July 13). AI in luence s a e aking o e adul con en . The Ve ge. h ps://www. he e ge.com
[29] Wagne , T. L., & Blewe , A. (2019). The wo d eal is no longe eal: Deep akes, gende , and he challenges o AI-
al e ed ideo. Open In o ma ion Science, 3(1), 32–41. h ps://doi.o g/10.1515/opis-2019-0003
