EUVIMEDEUVIMED entdecken
Crossref

Prompt Analysis in Large Language Models Adversarial Prompt Detection

M. Sriram, P. Vishvaas, M. Sai Sreejan, A. Sai Ramana, Mrs. K.B. Pullamma, Dr. K S R K Sarma

Research Digest on Engineering Management and Social Innovations · 23.04.2026 · 2 (4)

DOI: 10.46647/rdems0204018 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Research Digest on Engineering Management and Social Innovations, online erschienen am 23.04.2026. Verfasst von M. Sriram und 5 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Large Language Models such as GPT, LLaMA, and Claude have demonstrated remarkable capabilities in natural language understanding and generation, making them increasingly popular in applications such as chatbots, coding assistants, and enterprise solutions. However, the widespread adoption of these models has also raised serious concerns regarding their security, as adversarial prompt injection attacks, commonly known as “jailbreaking,” manipulate LLMs into generating harmful, biased or confidential outputs by bypassing safety measures. Traditional rule-based defenses are insufficient since adversarial prompts exploit the semantic and contextual nature of language. This project proposes the development of an AIdriven detection system to identify and block adversarial prompts before they reach the LLM by leveraging transformer-based NLP models (BERT, RoBERTa, LLaMA) along with adversarial training techniques to build a classifier that distinguishes between safe, suspicious and malicious prompts. The solution not only strengthens the security of LLM-based systems but also contributes to the emerging field of AI safety and robustness through experimental validation on standard adversarial NLP datasets and synthetic jailbreak attempts, targeting over 85% detection accuracy with explainable outputs that provide transparency into threat detection mechanisms.

Zitieren

APA

Sriram, M., Vishvaas, P., Sreejan, M. S., Ramana, A. S., Pullamma, M. K., & Sarma, D. K. S. R. K. (2026). Prompt Analysis in Large Language Models Adversarial Prompt Detection. Research Digest on Engineering Management and Social Innovations, 2(4). https://doi.org/10.46647/rdems0204018

Vancouver

Sriram M, Vishvaas P, Sreejan MS, Ramana AS, Pullamma MK, Sarma DKSRK. Prompt Analysis in Large Language Models Adversarial Prompt Detection. Research Digest on Engineering Management and Social Innovations. 2026;2(4). doi:10.46647/rdems0204018

BibTeX
@article{Sriram2026,
  author = {Sriram, M. and Vishvaas, P. and Sreejan, M. Sai and Ramana, A. Sai and Pullamma, Mrs. K.B. and Sarma, Dr. K S R K},
  title = {Prompt Analysis in Large Language Models Adversarial Prompt Detection},
  journal = {Research Digest on Engineering Management and Social Innovations},
  year = {2026},
  volume = {2},
  number = {4},
  doi = {10.46647/rdems0204018}
}
Kostenlos starten →