The increasing sophistication of phishing attacks necessitates the development of detection models that are not only accurate but also interpretable and secure. While Transformer-based models have shown great promise in this domain, their \u201cblack box\u201d nature often hinders trust and deployment in critical security environments. This paper presents a comprehensive, multi-faceted analysis of a DistilBERT-based model for phishing URL detection, moving beyond standard performance metrics to investigate its internal mechanisms, causal reasoning, and security posture. We first establish a performance benchmark by training a high-performance DistilBERT classifier that achieves a 95.85% F1-score after robust threshold optimization, significantly outperforming traditional machine learning baselines. To deconstruct the model’s reasoning, we introduce a domain-specific probing method to identify and analyze \u201cspecialist\u201d attention heads and contextualize these findings with established XAI methods like SHAP and LIME. A subsequent intervention-based ablation study on these heads yields a surprising insight into the model’s resilience: performance does not degrade when specialist heads are removed, revealing a high degree of distributed knowledge and functional redundancy. We evaluate theoretical adversarial vulnerability across multiple perturbation budgets (\u03f5\u2208{0.1, 0.2, 0.3, 0.5}), finding that the model exhibits strong resilience against small perturbations (F1 > 0.95 for \u03f5 \u2264 0.1) while showing graceful degradation under stronger attacks, consistent with security benefits arising from distributed representations. We further evaluate a battery of operationally realistic, non-gradient obfuscations\u2014including character transposition, leet substitution, homoglyph, and delimiter manipulations (e.g., micorsoft.com , paypa1-login.com )\u2014and find that evasion success never exceeds 7.2%, with the subword tokenizer rendering such manipulations more detectable rather than less. Our findings demonstrate that the model’s strength lies not in a few critical components but in a distributed, non-brittle reasoning process, making it a promising component within layered phishing defense systems.<\/p>\n","protected":false},"excerpt":{"rendered":"
The increasing sophistication of phishing attacks necessitates the development of detection models that are not only accurate but also interpretable and secure. While Transformer-based models have shown great promise in this domain, their \u201cblack box\u201d nature often hinders trust and deployment in critical security environments. This paper presents a comprehensive, multi-faceted analysis of a DistilBERT-based […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"text","value":"V. D","user_id":0},{"type":"text","value":"Geetha Dayalan","user_id":0},{"type":"user_nicename","value":"Abhay Bhandarkar","user_id":"44299"}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"Frontiers of Computer Science","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"Frontiers in Computer Science","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"10.3389\/fcomp.2026.1836418","msr_arxiv_id":"","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2026-08-31","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2026,"msr_month":8,"msr_day":31,"msr_microsoftintellectualproperty":false,"msr_pub_id":"f11dee894b3f0acbc4ac454e71911b470d91d46d","msr_publication_uploader":[{"type":"url","title":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/pdf","label_id":243132,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/full","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"f11dee894b3f0acbc4ac454e71911b470d91d46d","msr_s2_pdf_url":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/pdf","msr_citation_count_updated":"","msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":33,"msr_s2_open_access":true,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"f11dee894b3f0acbc4ac454e71911b470d91d46d"},{"provider":"doi","id":"10.3389\/fcomp.2026.1836418"},{"provider":"corpusid","id":"291636730"}],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13556,13558],"msr-publication-type":[193715],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[270657],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1189085","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-security-privacy-cryptography","msr-locale-en_us"],"msr_publishername":"","msr_edition":"Frontiers of Computer Science","msr_affiliation":"","msr_published_date":"2026-08-31","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"Frontiers in Computer Science","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"10.3389\/fcomp.2026.1836418","msr_publication_uploader":[{"type":"url","title":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/pdf","label_id":243132,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/full","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"f11dee894b3f0acbc4ac454e71911b470d91d46d","msr_influential_citations":0,"msr_reference_count":33,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":true,"msr_s2_pdf_url":"https:\/\/www.frontiersin.org\/journals\/computer-science\/articles\/10.3389\/fcomp.2026.1836418\/pdf","msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"V. D","user_id":0,"rest_url":false},{"type":"text","value":"Geetha Dayalan","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Abhay Bhandarkar","user_id":44299,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Abhay Bhandarkar"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"article","related_content":[],"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1189085","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":3,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1189085\/revisions"}],"predecessor-version":[{"id":1190610,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1189085\/revisions\/1190610"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1189085"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1189085"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1189085"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1189085"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1189085"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1189085"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1189085"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1189085"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1189085"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1189085"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1189085"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1189085"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1189085"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1189085"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}