Recent Trends in Artificial Intelligence & It’s Applications (e-ISSN: 2583-4819, p-ISSN: 3107-7234)

AI-Enabled Cyber Threat Intelligence Platform for Email Security: Detection, Geolocation, and Forensic Correlation

2026-10-09T09:15:07+00:00

Email remains one of the most widely used and trusted digital communication channels, which also makes it a prime target for cyberattacks. Adversaries exploit email through phishing campaigns, sender spoofing, malicious links, malware attachments, and social engineering to steal sensitive information or compromise systems. As these threats grow in volume and sophistication, manual inspection by security analysts becomes slow, error-prone, and unable to keep pace with modern attacks—creating a pressing need for automated, intelligent detection systems. To address this, this work proposes an AI-Powered Email Threat Detection, Geolocation, and Forensic Intelligence Platform that combines three core capabilities into one system: automated threat detection, geographic origin analysis, and digital forensic investigation. The platform uses AI and machine learning to analyze email characteristics such as header metadata, sender authentication signals, IP addresses, embedded URLs, and domain registration details. A classification model then labels each email as safe, suspicious, or malicious, allowing security teams to prioritize responses based on actual risk rather than manually reviewing every message. For emails flagged as suspicious or malicious, the platform performs geolocation analysis on sender IP addresses, mapping them to their likely geographic and network origin. This spatial context helps analysts identify attack patterns linked to specific regions, infrastructure providers, or known threat actors. The forensic layer further supports evidence correlation across incidents, helps trace the scope of larger campaigns, and maintains a documented trail for post-incident reporting. By integrating AI-driven detection, geolocation intelligence, and forensic analysis into a single platform, this system offers a faster, more comprehensive approach to identifying and investigating email-based threats—reducing analyst workload, improving threat awareness, and strengthening organizational resilience against evolving attack vectors.

Algorithmic and AI-augmented Human Resource Management: A Review of Recruitment, Predictive Attrition Analytics, People Analytics, and Algorithmic Management Mechanisms

2026-09-17T08:09:13+00:00

Organizations increasingly delegate recruitment screening, attrition forecasting, workforce planning, and even task allocation to algorithmic and machine-learning systems, a shift that has outpaced the development of a unified evidence base comparing these mechanisms on efficiency, fairness, and employee-experience grounds. This review synthesizes peer-reviewed and conference literature on four major families of algorithmic Human Resource Management (HRM) practice — algorithmic recruitment and screening, predictive employee-attrition analytics, descriptive-to-prescriptive people analytics, and algorithmic management on gig and platform work — and compares them on reported efficiency gains, fairness risk, and organizational maturity. A structured narrative review was conducted across IEEE Xplore, SpringerLink, Elsevier ScienceDirect, Emerald, Wiley, and Nature Portfolio journals, screening peer-reviewed studies reporting empirical or conceptual evidence on algorithmic HRM mechanisms published between 2005 and 2026. 33 sources are synthesized into comparative tables spanning mechanism, reported outcome, and principal limitation, alongside six illustrative computed figures benchmarking adoption trajectories, attrition-classifier learning behavior, efficiency-versus-fairness-risk positioning, and reported milestones over 2011–2025. No single reviewed mechanism simultaneously maximizes efficiency, minimizes fairness risk, and sustains employee trust; predictive attrition analytics and maturity-staged people analytics currently offer the most favorable balance, while algorithmic recruitment and gig-platform algorithmic management report the largest efficiency gains alongside the highest documented fairness and governance risk, identifying human-centered algorithmic governance as the central open problem for the field.

Artificial Intelligence and Continuing Professional Development: A Review of Emerging Applications

2026-09-15T11:49:29+00:00

Continuing Professional Development (CPD) is no longer a routine formality; it is the primary means by which professionals remain current, competent, and responsive to continuous change in technology and the workplace. Artificial Intelligence (AI) is reshaping this landscape by making learning more personalized, intelligent, and adaptive. AI-enabled systems offer tailored training, intelligent assessments, flexible content delivery, and decisions grounded in robust data. This review examines the application of AI in CPD across diverse fields, including education, engineering, healthcare, and management. Drawing on recent studies, it identifies the principal AI tools in use, examines implementation approaches, and evaluates outcomes, challenges, and future directions. The findings indicate that AI enhances learning efficiency, increases engagement, broadens access, and improves professional development outcomes. At the same time, significant challenges remain, including privacy concerns, ethical considerations, algorithmic bias, and gaps in digital literacy. The paper argues that, when implemented thoughtfully, AI has the potential to substantially transform professional learning and development for the better.

Assisting Blind Individuals in Navigation Using Multimodal LLM and Voice Assistance

2026-09-03T04:59:17+00:00

Visual impairment can considerably limit a person’s ability to move through unfamiliar surroundings safely and independently. Recent developments in Artificial Intelligence (AI), Computer Vision, Deep Learning, Multimodal Large Language Models (LLMs), and smartphone-based sensing technologies have facilitated the emergence of intelligent assistive navigation systems designed to support safer and more independent mobility. These systems employ object detection, distance estimation, scene understanding, voice guidance, and haptic feedback to improve environmental awareness and obstacle avoidance. This survey reviews recent research on AI-based assistive navigation systems for visually impaired individuals, with particular emphasis on object detection models, multimodal LLMs, LiDAR-based sensing, distance estimation, and real-time audio guidance. The reviewed approaches demonstrate improvements in object recognition, environmental understanding, and navigation assistance; however, several limitations remain, including dependence on specialized hardware, high computational requirements, response latency, limited offline functionality, and insufficient integration of multiple assistive features within a single application. The survey identifies these research gaps and discusses future directions toward lightweight, affordable, smartphone-based assistive navigation solutions. Integration of multimodal AI, efficient object detection, smartphone sensors, real-time distance estimation, voice assistance, vibration feedback, emergency support, and offline processing can contribute to safer, more reliable, and accessible navigation for visually impaired individuals.

Transition from Chatbots to Computer-Using AI Agents: A Systematic Review of Autonomous Workflows, Evaluation, Security, and Human Oversight

2026-10-17T06:06:07+00:00

These are enormous language models (LLMs) that enable not only the generation of information, but also acting within external software environments through tools, browsers, application interfaces, code execution, and computer-use mechanisms. This is a qualitative change in behaviour, introducing an interesting research problem since the ability to make a believable response goes hand-in-hand with creating, mutating, or deleting external state. This systematic review explores the required transition from conversational LLMs to computer-using AI agents, and studies autonomous workflows, evaluation, safety, trustworthiness, and human oversight. B. Role of background on reasoning-and-acting and tool use with evidence from agent, web, desktop, software-engineering, and safety benchmarks (vs React/Tool former vs OSWorld 2.0/swe-bench/AgentDojo w AI-comp-towards-24k/big-k-c) using-agent safety and security. This review uses a candidate eligibility protocol, a screening log at the source level, explicit pre-defined analytical categories, and systematic data extraction along categories including environment, action authority, planning horizon in space and time, verification of execution from intent to movement, as well as efficiency, safety, and human oversight. This synthesis demonstrates that agentic capability consists of multiple dimensions. Simply completing tasks is not enough, as an agent could arrive at a correct final state performing irrelevant actions, breaking through authorization boundaries, losing track of concealed state, or executing commands in untrusted content. Realistic interaction still proves much more difficult than isolated question answering, and long-horizon evaluation reveals additional limitations in constraint tracking, cross-source reasoning, recovery, and verification, as shown by web and desktop benchmarks. A security study shows that granting access to the tools is opening completely new attack surfaces that cannot be covered by model-level refusal behavior. The review thus identifies five interacting deployment dimensions: capability, environmental access, action autonomy, monitoring and verification, and human oversight. They are organized in a proposed Human-Oversight and Agentic Control (HOAC) conceptual framework, not yet a validated quantitative model for empirical evaluation. An existing frontier-model case study, GPT-6 Astra (LLM), is presented separately based on its academic evidence, as its benchmark and safety claims are developer-reported and therefore do not constitute independent validation. The review also concludes: permission-aware architectures, long-horizon evaluation, independent benchmarking, stronger defences against prompt injection and Bias amplification, risk-based human intervention, and interoperable audit trails. Computer-using agents should be evaluated as socio-technical control systems where having the capability of modelling is only one component of safety and reliability.

Human or Machine? A Random Forest–Based Framework for Detecting AI-Generated Speech Using Acoustic Feature Analysis

2026-10-08T11:39:56+00:00

The rapid advancement of AI-based voice synthesis technologies has made it increasingly difficult to distinguish authentic human speech from artificially generated audio, raising concerns around misinformation, fraud, and digital security. This study presents an AI-Based Voice Detection system, Voicelytics, that analyzes uploaded audio or video files and classifies the contained voice as human or AI-generated using a Random Forest classifier trained on extracted acoustic features. The system extracts Mel-Frequency Cepstral Coefficients (MFCC), spectral centroid, spectral rolloff, zero-crossing rate, RMS energy, and pitch from input audio, applies feature scaling, and generates a classification with an associated confidence score through a Flask-based web interface. To rigorously evaluate this approach, a Random Forest classifier was trained and tested on the DEEP-VOICE benchmark dataset — 11,778 balanced samples of real human speech and Retrieval-based Voice Conversion (RVC) deepfakes from eight public figures — using an 80/20 stratified train-test split. On the held-out test set, the classifier achieved 98.73% accuracy, 98.24% precision, 99.24% recall, a 98.73% F1-score, a 1.10% Equal Error Rate (EER), and a ROC-AUC of 0.9993. These results are consistent with, and slightly exceed, the 90–98% accuracy range reported by comparable MFCC-plus-Random-Forest studies in the literature. The system is positioned against prior studies spanning perceptual deep learning, transformer-based, ensemble, benchmark-dataset, and classical feature-based approaches to deepfake speech detection. This work contributes an accessible, locally deployable, interpretable, and empirically validated tool for voice authenticity verification in digital communication, media, and security contexts.

Weekend Genie: An AI-Powered Personalized Travel Itinerary Generation System Using Large Language Models

2026-09-15T12:09:08+00:00

Automated travel planning using Large Language Models (LLMs) is a more dynamic and innovative solution than conventional online travel agencies and manual, time-consuming curation, offering a more intelligent personal assistant. This paper presents an end-to-end Artificial Intelligence-based system for generating a travel itinerary, Weekend Genie, for a weekend trip to various travel destinations in India using LLaMA 3.3 70B Versatile with Groq base inference engine, which generates a highly structured, context-rich and cost-effective travel itinerary. Once simple parameters like total budget, group size, origin, and travel dates have been entered, the user gets a complete multi-day itinerary, which includes optimized recommendations for accommodation and meals, based on the origin of the group, cultural offers, and hidden spots. Architecturally, Weekend Genie has a modern decoupled software stack that couples the front-end, in JavaScript (React 18 and TypeScript), a backend, in a RESTful API (Express.js 5), and a state management mechanism (Redux) with a JSON-schemas (generated from an engineered prompt) for session security and constrained data, all of which are backed by a MongoDB (document) store for user profiles, stored trip histories, etc. The generation latency from end to end is 3-8 seconds, and 89% of all the experiments follow the JSON schema, with 97% enforcing budget compliance. Weekend Genie also performs better than other commercial and academic AI-based travel planning tools on four metrics: structural results, local relevance and reliability, price accuracy, and travel itinerary generation. All of the code is open-sourced so that it can be used as a production-ready reference implementation for applications that use LLMs.

Arogyabodhini: An AI-Powered Multilingual Healthcare Assistant for Disease Prediction and Smart Medical Support

2026-09-05T12:04:25+00:00

Healthcare accessibility remains a major challenge, particularly in multilingual regions where patients often struggle to communicate their symptoms due to language barriers, limited medical knowledge, and restricted access to specialized healthcare services. Existing digital healthcare systems generally support a limited number of languages and lack an integrated platform for multilingual communication, intelligent symptom analysis, doctor recommendations, and medical report processing. To address these challenges, this article proposes Arogyabodhini, an AI-powered multilingual healthcare assistant that provides intelligent medical support through voice-enabled interaction, disease prediction, and smart healthcare services. The proposed system enables patients to describe their health conditions using voice or text in regional Indian languages, where speech input is converted into a professional language (English) using speech recognition and translation techniques to ensure standardized processing. Natural Language Processing (NLP) is employed to extract relevant symptoms from patient descriptions, which are then analyzed using a Decision Tree-based machine learning model to predict possible diseases. Furthermore, Arogyabodhini integrates a centralized doctor recommendation system that identifies suitable specialists based on predicted diseases, specialization, and availability while supporting online appointment booking, digital prescription management, and emergency alerts for critical health conditions. A secure and user-friendly dashboard enables patients to access their health records, consultation history, uploaded reports, and medical recommendations. By integrating Artificial Intelligence, Machine Learning, Natural Language Processing, speech recognition, and modern web technologies into a unified healthcare platform, the proposed system aims to improve healthcare accessibility, reduce communication barriers, support early disease identification, and simplify patient–doctor interaction, thereby contributing to the development of inclusive, intelligent, and technology-driven healthcare services, particularly for rural and linguistically diverse communities.

Neuromorphic Computing to make AI Energy-efficient: A Review of Contemporary Architectures, Difficulties and Applications

2026-08-31T07:24:09+00:00

Artificial intelligence is placing increasingly demanding requirements, especially in data-intensive and real-time applications, on the efficiency and scalability of traditional Von Neumann computing architectures, hence causing huge energy demands and computational bottlenecks. This article gives a detailed overview of neuromorphic computing, which is a relatively new paradigm that finds its inspiration in highly efficient systems and provides a potential solution to these limitations. The work defines the foundational principles of neuromorphic systems, such as the details of Spiking Neural Networks (SNNs), the benefits of event-driven processing, and the synaptic plasticity processes, which, together, will allow never-before-seen energy efficiency to be achieved. It gives a more detailed analysis of various hardware implementations that include digital and analog neuromorphic chips, etc. In addition, the review also discusses the wide range of applications in which neuromorphic systems outperform, such as real-time edge AI and autonomous robotics. The major achievements are the description of modern neuromorphic hardware and an overview of all existing and potential applications.