Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring
Thesis (MA)--Stellenbosch University, 2025.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Stellenbosch : Stellenbosch University
2025
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614004978909184 |
|---|---|
| access_status_str | Open Access |
| author | Moabi, Boitshwarelo Precious |
| author2 | Blaauw, D. N. |
| author_browse | Blaauw, D. N. Moabi, Boitshwarelo Precious |
| author_facet | Blaauw, D. N. Moabi, Boitshwarelo Precious |
| author_sort | Moabi, Boitshwarelo Precious |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MA)--Stellenbosch University, 2025. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/134707 |
| institution | Stellenbosch University (South Africa) |
| language | English |
| last_indexed | 2026-06-10T12:45:08.467Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/134707 Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring Moabi, Boitshwarelo Precious Blaauw, D. N. Stellenbosch University. Faculty of Arts and Social Sciences. Dept. of Information Science. Artificial intelligence -- Security measures Computer security -- Data processing Computer security -- Risk assessment Computer security -- Decision making Artificial intelligence -- Moral and ethical aspects Machine learning -- Social aspects UCTD Thesis (MA)--Stellenbosch University, 2025. Moabi, B. P. 2025. Exploring Algorithmic Bias in Cybersecurity: The Social Implications of AI-Driven Threat Detection and Risk Scoring. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/ad3878d5-ef9f-4421-b6f8-54910429a38f ENGLISH SUMMARY: The reliance on Artificial Intelligence (AI) in cybersecurity has expanded and revolutionized how threats are detected and risks assessed. However, critical challenges are introduced in this advancement; algorithmic bias which can reproduce existing social and inequalities unintentionally. This study explores the manifestations, causes and implications of algorithmic bias in AI-driven cybersecurity systems, with specific focus on both threat detection and risk scoring mechanisms. Guided by an interpretivist paradigm, the study employed a qualitative approach utilizing semi-structured interviews with cybersecurity professionals who possess practical experience in deploying, as well as managing artificial intelligence-based security systems and models. Data were analysed inductively through thematic analysis to capture patterns of meaning regarding perceptions, ethical concerns and challenges surrounding algorithmic bias in cybersecurity decision making Key findings indicates that biases often stem from skewed pattern selection and unrepresentative training datasets, lack of transparency in artificial intelligence decision process and model assumptions. These biases lead to false positives and false negatives in risks identification, thereby compromising accountability, fairness and the transparency of AI-driven cybersecurity systems. The results further reveal that mitigating bias requires a strategic approach from multi-level interventions including incorporation of bias audits, adoption of transparent AI models, enhanced data diversity and human oversight in automated threat detection systems. The research contributes to the emerging topic of AI ethics in cybersecurity by offering empirical evidence on how algorithmic bias impact the reliability of technology and social trust in digital security ecosystems. Thus, providing a conceptual foundation for formulating ethical, equitable and explainable AI frameworks that can aid both practitioners and policymakers in designing bias awareness cybersecurity systems. AFRIKAANSE OPSOMMING: Die afhanklikheid van Kunsmatige Intelligensie (KI) in kuberveiligheid het uitgebrei en 'n revolusionering teweeggebring in hoe bedreigings opgespoor en risiko's beoordeel word. Hierdie vooruitgang bied egter kritieke uitdagings. Algoritmiese vooroordeel wat bestaande sosiale en ongelykhede onbedoeld kan reproduseer. Hierdie studie ondersoek die manifestasies, oorsake en implikasies van algoritmiese vooroordeel in KI-gedrewe kuberveiligheidstelsels, met spesifieke fokus op beide bedreigingsopsporing en risikobepalingsmeganismes. Gelei deur 'n interpretiwistiese paradigma, het die studie 'n kwalitatiewe benadering gebruik deur semi-gestruktureerde onderhoude met kuberveiligheidspesialiste te gebruik wat praktiese ervaring het in die ontplooiing, sowel as die bestuur van kunsmatige intelligensie-gebaseerde sekuriteitstelsels en -modelle. Data is induktief deur middel van tematiese analise geanaliseer om betekenispatrone rakende persepsies, etiese bekommernisse en uitdagings rondom algoritmiese vooroordeel in kuberveiligheidsbesluitneming vas te le. Belangrike bevindinge dui daarop dat vooroordele dikwels voortspruit uit skewe patroonseleksie en onverteenwoordigende opleidingsdatastelle, 'n gebrek aan deursigtigheid in die kunsmatige intelligensie-besluitnemingsproses en modelaannames. Hierdie vooroordele lei tot vals positiewe en vals negatiewe in risiko-identifikasie, wat waardeur aanspreeklikheid, billikheid en die deursigtigheid van KI-gedrewe kuberveiligheidstelsels in die gedrang kom. Die resultate toon verder dat die vermindering van vooroordeel 'n strategiese benadering vereis vanaf multivlak-intervensies, insluitend die insluiting van vooroordeeloudits, die aanvaarding van deursigtige KI-modelle, verbeterde datadiversiteit en menslike toesig in outomatiese bedreigingsopsporingstelsels. Die navorsing dra by tot die opkomende onderwerp van KI-etiek in kuberveiligheid deur empiriese bewyse te bied oor hoe algoritmiese vooroordeel die betroubaarheid van tegnologie en sosiale vertroue in digitale sekuriteitsekosisteme beinvloed. Dit bied dus 'n konseptuele grondslag vir die formulering van etiese, billike en verklaarbare KI-raamwerke wat beide praktisyns en beleidmakers kan help om vooroordeelbewustheid-kuberveiligheidstelsels te ontwerp. Masters 2025-12-24T09:24:24Z 2025-12-24T09:24:24Z 2025-12 Thesis https://scholar.sun.ac.za/handle/10019.1/134707 en Stellenbosch University xi, 136 pages : illustrations, includes annexures application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Artificial intelligence -- Security measures Computer security -- Data processing Computer security -- Risk assessment Computer security -- Decision making Artificial intelligence -- Moral and ethical aspects Machine learning -- Social aspects UCTD Moabi, Boitshwarelo Precious Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title | Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title_full | Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title_fullStr | Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title_full_unstemmed | Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title_short | Exploring algorithmic bias in cybersecurity : the social implications of AI-driven threat detection and risk scoring |
| title_sort | exploring algorithmic bias in cybersecurity the social implications of ai driven threat detection and risk scoring |
| topic | Artificial intelligence -- Security measures Computer security -- Data processing Computer security -- Risk assessment Computer security -- Decision making Artificial intelligence -- Moral and ethical aspects Machine learning -- Social aspects UCTD |
| url | https://scholar.sun.ac.za/handle/10019.1/134707 |
| work_keys_str_mv | AT moabiboitshwareloprecious exploringalgorithmicbiasincybersecuritythesocialimplicationsofaidriventhreatdetectionandriskscoring |