Developing Smart/Generative AI models to enhance the safeguards in securing cloud AI workloads from adversarial attacks

Rajvansh Chaudhary

Developing Smart/Generative AI models to enhance the safeguards in securing cloud AI workloads from adversarial attacks

The proliferation of generative AI workloads hosted on cloud platforms introduces novel adversarial threat vectors—including evasion, data poisoning, prompt injection, and model stealing—that jeopardize model integrity, confidentiality, and availability. This paper surveys adversarial attacks targeting generative models (e.g., GANs, VAEs, LLMs) and cloud-specific vulnerabilities, then examines defense mechanisms such as adversarial training, input sanitization, federated detection, access control, and prompt filtering. Comparative analyses assess each method’s effectiveness, overhead, and adaptability. We highlight challenges unique to cloud deployment—multi-tenancy, dynamic scaling, shared infrastructure—and propose a layered, design for security approach integrating detection, robust design, and governance. The future roadmap underscores auditability, explainability, and collaboration across cloud providers. The paper synthesizes literature spanning 2013–2023, offering a strategic foundation for securing cloud based generative AI systems.

Published Paper   E-Certificate

"Developing Smart/Generative AI models to enhance the safeguards in securing cloud AI workloads from adversarial attacks", IJSDR - International Journal of Scientific Development and Research (www.IJSDR.org), ISSN:2455-2631, Vol.9, Issue 5, page no.1576-1582, May-2025, Available :https://ijsdr.org/papers/IJSDR2405208.pdf

Volume 9 Issue 5, May-2025

Pages : 1576-1582

Paper Reg. ID: IJSDR_304943

Published Paper Id: IJSDR2405208

Downloads: 000138

Research Area: Science and Technology

Country: -, -, India

Published Paper PDF: https://ijsdr.org/papers/IJSDR2405208

Published Paper URL: https://ijsdr.org/viewpaperforall?paper=IJSDR2405208

ISSN: 2455-2631 | IMPACT FACTOR: 9.15 Calculated By Google Scholar | ESTD YEAR: 2016

An International Scholarly Open Access Journal, Peer-Reviewed, Refereed Journal Impact Factor 9.15 Calculate by Google Scholar and Semantic Scholar | AI-Powered Research Tool, Multidisciplinary, Monthly, Multilanguage Journal Indexing in All Major Database & Metadata, Citation Generator

Publisher: IJSDR(IJ Publication) Janvi Wave

This work is licensed under a Creative Commons Attribution 4.0 International License and The Open Definition