Procter & Gamble (United Kingdom)
RhinoSec - Repository of the Faculty of Security StudiesNot a member yet
488 research outputs found
Sort by
Cyber threats and energy security: Development and analysis of an incident dataset for the period 2022–2024
No full textDriven by technological progress, cyber threats present one of the most prominent security challenges in the modern world. The ongoing transformation of energy sector is creating increasingly interconnected energy systems that rely on digital technologies. As a vital component of the economy, the energy sector is highly exposed to numerous simultaneous threats, including cyber-attacks, where the motivations behind such incidents are difficult to understand and may range from financial gains to politically motivated targeting. By developing highly resilient components within the energy system, a country can strengthen its wider concept of energy security, protecting the critical infrastructure and continuity of energy supply. This paper aims to provide a sector analysis for a selected three-year period, from 2022 to 2024, applying a thorough research approach to create a broader picture of the current trend. The dataset was developed, reflecting the sector’s vulnerability. As a result, major events are presented, indicating the scope of potential threats, their growing complexity and sophistication, and their eventual consequences. The paper highlights the importance of the cyber component within the energy security concept, one which is often insufficiently considered in both theoretical and practical terms. Accordingly, this work aims to provide evidence-based support for an alternative energy security theo
retical framework combining security of critical infrastructure and security of supply principles.https://www.sciencedirect.com/science/article/pii/S030142152500420
AI incidents and data integrity
No full textArtificial intelligence (AI) has experienced widespread adoption across diverse sectors due to its capacity to enhance operational efficiency and economic competitiveness. However, the deployment of AI systems has also simultaneously introduced numerous security challenges and potential risks which demand careful consideration. As investments in AI development have increased substantially, corresponding investments in cybersecurity have become more critical. Ensuring the secure implementation of AI, particularly within critical infrastructure systems, necessitates the development of robust and resilient systems. Analysing real world AI incidents provides valuable insights which may serve to enhance security mechanisms and prevent future vulnerabilities. Since the inception of artificial intelligence, researchers have identified various system vulnerabilities and associated risks. Promoting awareness of such potential AI hazards has proven instrumental in facilitating a deeper understanding of both the scope and severity of these risks. Moreover, such awareness provides a framework for developing AI tools which are not only more resilient, but also ethically sound.
It is essential to examine how these dynamics evolve in practical environments by systematically identifying such incidents, their underlying causes, and their consequent impacts, rather than relying solely on theoretical projections. In response to this urgent need, AI incident databases have emerged as crucial instruments for responsible AI development and governance. The primary objective of these initiatives is to methodically document and categorise incidents, thereby strengthening security measures, informing preventative strategies, and fostering transparency and accountability within AI systems management.https://www.mi.sanu.ac.rs/~ai_conf
Između polova: Uticaj polarnosti na opseg strateških opcija malih država
No full textPolazeći od Tukididovog zapažanja da snažni čine ono
što mogu, a slabi trpe ono što moraju, u radu se razmatraju načini na
koje različiti oblici polarnosti međunarodne strukture utiču na
opseg strateških opcija malih država. Tragajući za pružanjem
generalizovanog objašnjenja uticaja polarnosti na strateške opcije
malih država, centralna teza rada jeste da različiti oblici
polarnosti omogućavaju različit stepen slobode delanja država. Sa
povećanjem broja polova, povećava se i sloboda delanja i opseg
strateških opcija malih država usled dva ključna razloga. Prvi je
što prisustvo više moćnih aktera ograničava jednostrano delovanje,
pošto velike sile moraju da uzimaju u obzir reakcije drugih sila. Ovo
proizvodi opreznije ponašanje i manju sklonost ka agresiji prema
slabijima. Drugi je što veći broj velikih sila otvara nove prilike
malim državama za delanje zbog mogućnosti implementacije strategija
koje počivaju na trijadnim interakcijama. U radu se zaključuje da iako
su materijalne sposobnosti važne, u multipolarnim sistemima
umešno državništvo i dobra strategija mogu znatno povećati
mogućnost malih država da ostvare šire interese od pukog
preživljavanja. Na kraju, u radu se iskazuje i oprez da iako
multipolarni sistemi omogućavaju veću fleksibilnost i prostor za
aktivnu spoljnu politiku, oni nose veću opasnost po stabilnost i
opstanak, što je naročit slučaj u trenutnom momentu kada je sistem u
strukturalnom fluksu
Moguće opasnosti generativne veštačke inteligencije
No full textGenerativna veštačka inteligencija (GenAI) predstavlja oblast proučavanja koja obuhvata razvoj velikih modela sa milijardama parametara, omogućavajući generisanje sadržaja u različitim medijima. Zahvaljujući svojoj sposobnosti da unapredi efikasnost i ekonomsku konkurentnost GenAI je već našla široku primenu u brojnim oblastima, kao što su bankarstvo, zdravstvo, biologija, saobraćaj, visoko obrazovanje, energetika i druge. Međutim, pored brojnih prednosti, GenAI istovremeno nosi i značajne bezbednosne rizike. Primeri zloupotrebe sistema GenAIobuhvataju kreiranje dezinformacija, organizovanje ciljanih fišing kampanjai druge manipulativne aktivnosti. Pored toga, sofisticirani napadi poput suparničkih, ubacivanja malicioznih unosa (prompt injection), ekstrakcije podataka i inverzije modela ciljaju ranjivosti GenAI sistema kako bi manipulisali korisnicima ili ostvarili finansijsku korist. Upravo zbog ovih izazova, neophodno je temeljno analizirati potencijalne rizike u stvarnim scenarijima i istražiti odgovarajuće strategije ublažavanja, čime bi se obezbedila bezbedna i etička primena GenAI u oblastima od ključno
Osporeni suverenitet: Kopnena zona bezbednosti kao geopolitičko pograničje
No full textU radu se posmatra prostor Kopnene zone bezbednosti (KZB-a) iz geopolitičke perspektive. Iako je potpisivanjem vojno-tehničkog sporazuma u Kumanovu, Srbija prepustila međunarodnoj zajednici faktičku kontrolu nad prostorom Autonomne pokrajine Kosova i Metohije (AP KiM), i time „osporila“ sopstveni suverenitet, isto se može reći i za prostor KZB-a. Na ovom području suverenitet Srbije je takođe bio osporen, ali implikacije osporenog suvereniteta nisu bile detaljnije proučavane. Centralna teza ovog rada jeste da je osporeni suverenitet postojao i da na neki način i dalje postoji i na prostoru KZB-a, s obzirom da Zona predstavlja svojevrsno geopolitičko pograničje (borderlands). Srž osporenog suvereniteta u KZB-u proizilazi iz činjenice da je reč o geopolitičkom pograničju koji počiva na hibridnom bezbednosnom poretku u kojem je ipak postojao svojevrsni bezbednosni vakuum. Taj vakuum nije proizvod nemogućnosti za njenom kontrolom već specifičnog pravnog poretka nametnutog uslovima međudržavnog tehničkog sporazuma kojim je zaustavljena NATO agresija na tadašnju SR Jugoslaviju
Disinformation and generative AI: risks, chalenges and possible solutions
No full textThe disinformation phenomenon has been present since the very dawn of human communication, with repeated exposure to false information often resulting in its acceptance as truth. In contemporary society, however, disinformation has become increasingly alarming due to the simple, inexpensive, and convincing way it can be created using generative artificial intelligence and large language models in particular. These models, which continue to make significant advancements, enable the generation of large volumes of content which appear credible, even in the national language. In addition, such models facilitate the creation of personalised content tailored to certain groups or individuals, citing seemingly credible sources, and thus serving to further strengthen the impact of disinformation. In a world inundated with information, this problem has become even more pronounced. One of the key channels for the spread of disinformation is social media, whose primary goal is to capture and maintain their users' attention with content which reinforces their existing beliefs. This paper will present the potential uses of generative AI in the creation of disinformation, as well as possible solutions to address this problem.Fenomen dezinformacija prisutan je od samih početaka ljudske komunikacije, pri čemu
ponovljena izloženost lažnim informacijama često dovodi do njihovog prihvatanja kao istine. U
savremenom društvu, međutim, dezinformacije postaju sve zabrinjavajuće zbog jednostavnog,
jeftinog i uverljivog načina na koji se mogu kreirati koristeći generativnu veštačku inteligenciju,
naročito velike jezičke modele. Generartna veštčka inteligencija, koji i dalje prave značajne
napretke, omogućavaju generisanje velikih količina sadržaja koji izgledaju verodostojno, čak i na
nacionalnom jeziku. Pored toga, ovi modeli olakšavaju kreiranje personalizovanog sadržaja
prilagođenog određenim grupama ili pojedincima, navodeći prividno pouzdane izvore, čime
dodatno jačaju uticaj dezinformacija. U svetu preplavljenom informacijama, ovaj problem
postaje još izraženiji. Jedan od ključnih kanala za širenje dezinformacija su socijalni mediji, čiji
je primarni cilj da privuku i zadrže pažnju korisnika sadržajem koji potvrđuje njihova postojeća
uverenja. U radu je predstvljen potencijal upotrebe generativne veštačke inteligencije u kreiranju
dezinformacija, kao i mogućnosti za suzbijanje ovog problema.preprin
Fake News and Generative Artificial Intelligence: Risks and Potential Solutions
No full textThe fake news phenomenon has been present since the very dawn of human communication,
with repeated exposure to false information often resulting in its acceptance as truth. In
contemporary society, however, fake news has become extremely dangerous due to the simple,
inexpensive, and convincing way it can be created using generative artificial intelligence, and
large language models in particular. These models, which continue to make significant
advancements, enable the generation of large volumes of content which appear credible, even
in the Serbian language. In addition, such models facilitate the creation of personalised content
tailored to certain groups or individuals, citing seemingly credible sources, and thus serving to
further strengthen the impact of fake news. In a world inundated with information, this problem
has become even more pronounced. One of the key channels for the spread of fake news is
social media, whose primary goal is to capture and maintain their users' attention with content
which reinforces their existing beliefs. This process results in an echo chamber effect, where
users receive information which in turn reinforces their preconceptions, making it harder to
recognise false information. A secondary, but equally concerning issue caused by fake news is
the creation of distrust and confusion among users, ultimately leading to scepticism towards
even accurate information. The vast amount of data on social media highlights the need for the
application of artificial intelligence, not only to deliver relevant content, but also to detect fake
news. Although AI-based approaches can be used to identify false information, questions
remain as to the transparency and reliability of these algorithms in carrying out this task. This
paper will present the potential uses of artificial intelligence in the creation of fake news, as
well as possible solutions to address this problem.https://judig.jerteh.rs
The Impact of Gender, Seniority, Knowledge and Interest on Attitudes to Artificial Intelligence
No full textArtificial intelligence (AI) has become deeply rooted in our lives, yet uncertainties persist
regarding public attitudes to it, particularly among young individuals poised to engage with AI in their future careers. Understanding their perspectives is crucial not only for shaping educational frameworks, but also for assessing students’ readiness to navigate the rapidly evolving technological landscape in the modern workspace. This paper examines students’ attitudes to AI, as well as their interest in and knowledge about it. An adapted version of the Pew Research Center survey was used in our study to explore how gender and student seniority influence attitudes to AI generally and in specific applications such as facial recognition and self-driving cars. Our aim was to test the effects of these factors on AI attitudes, and to discover how various factors such as socio-demographics, knowledge, and interest may individually or collectively impact on AI attitudes in general, as well as in specific areas such as self-driving cars, facial recognition, or social media algorithms for fake news. We also investigated whether knowledge of AI and interest in it may serve to predict attitudes beyond the impacts of student seniority and gender. Our findings indicate that males self-report greater interest than females, but similar knowledge and general attitudes as female participants. Senior students possessed more AI knowledge compared to freshmen, but similar attitudes towards AI in general and self-driving cars. Interest in AI emerged as a significant predictor of general attitudes to AI and to self-driving cars, suggesting that increased interest correlates with more positive attitudes
Školska klima kao faktor bezbednosti u srednjim školama
No full textU radu se ispituje povezanost između školske klime i bezbednosti u srednjoj školi. Pozitivna školska klima i bezbedna školska sredina predstavljaju značajne faktore akademskog uspeha i pravilnog psihosocijalnog razvoja učenika. Stoga je važno ispitati strukturu i međusobnu povezanost indikatora školske klime i bezbednosti u školi kako bi se doprinelo jasnijem i utemeljenom kreiranju školskih politika. Istraživanje je sprovedeno putem upitnika. Upitnik školske klime meri tri faktora: podršku nastavnika i škole, učeničko zalaganje i interakcije, i strogost discipline. Upitnik bezbednosti u školi meri četiri faktora: nasilno ponašanje, (ne)bezbednost školske sredine, viktimizaciju i kršenje normi. U istraživanju je učestvovalo 719 učenika iz devet srednjih škola na teritoriji Republike Srbije. Istraživanjem je potvrđena povezanost između školske klime i bezbednosti u školi. Konkretno, podrška nastavnika i škole gradi značajne niske negativne povezanosti sa svim dimenzijama bezbednosti u školi. Učeničko zalaganje i interakcije grade srednju negativnu povezanost sa kršenjem normi, dok sa druge tri dimenzije bezbednosti u školi gradi niske negativne povezanosti. Strogost discipline treba posmatrati kao zasebnu pojavu u odnosu na koncept školske klime, a ona gradi niske pozitivne povezanosti sa kršenjem normi i nebezbednom školskom sredinom. Promovisanje pozitivnih interakcija među svim učesnicima obrazovno-vaspitnog procesa putem različitih aktivnosti podsticanja i razvoja poželjnih oblika ponašanja (poučavanje o ispravnim načinima rešavanja konflikata, podrška u obrazovnom procesu itd.) od strane škole i nastavnika znatno bi doprinelo izgradnji bezbedne školske sredine