TY  - JOUR
AU  - G Raju
AU  - Panga Sravani
AU  - Gudisili Suresh
AU  - Shaik Sameera Jasmine
AU  - Kamurthi Vardhan
AU  - Chinthalarevu Supriya
PY  - 2026
DA  - 2026/04/25
TI  - Conversational AI Based Assistant For Detecting  And Responding To Phishing Attacks
JO  - Global Journal of Engineering Innovations and Interdisciplinary Research
VL  - 6
IS  - 4
AB  - Phishing attacks have become increasingly sophisticated, exploiting human vulnerabilities through 
social engineering techniques such as urgency, authority, impersonation, and fear to deceive users into 
revealing sensitive information. Conventional phishing detection mechanisms, including blacklist-based 
filters and rule-driven heuristic systems, depend largely on predefined signatures, known malicious 
patterns, and surface-level technical indicators. While effective against previously identified threats, 
these approaches struggle to detect zero-day attacks, obfuscated URLs, context-aware impersonation, 
and emerging threats such as QR-code–based phishing, commonly referred to as quishing, and they 
offer limited transparency or user awareness. To overcome these limitations, this work proposes 
a conversational AI–based, human-centric assistant that actively supports users in detecting, 
understanding, and responding to phishing attacks through interactive dialogue. The system leverages 
Large Language Models integrated via LangChain to analyze phishing attempts across multiple vectors, 
including email text, embedded hyperlinks, and QR codes, with a strong emphasis on semantic and 
contextual analysis rather than simple pattern matching. Instead of producing a binary classification, 
the assistant identifies and explains key phishing indicators such as psychological manipulation cues, 
urgency triggers, authority abuse, and threat-based consequences using clear natural language. In 
addition, the system prioritizes data privacy through local processing and incorporates secure user 
authentication, administrative control, and OTP-based account recovery mechanisms. Experimental 
observations indicate that the proposed approach adapts effectively to evolving phishing strategies, 
enhances user awareness, and improves decision-making, demonstrating its potential as an explainable, 
adaptive, and practical AI-driven solution for addressing modern cybersecurity challenges in real-world 
environments across diverse digital communication platforms and user environments.
SN  - 3066-1226
UR  - https://dx.doi.org/10.33425/3066-1226.1298
DO  - 10.33425/3066-1226.1298