paperJuly 2025Unreviewed

QSAF: A Novel Mitigation Framework for Cognitive Degradation in Agentic AI

Hammad Atta, M. Baig, Yasir Mehmood, Nadeem Shahzad, Ken Huang, M. A. U. Haq, Muhammad Awais, Kamal Ahmed

arXiv.org

Abstract

We introduce Cognitive Degradation as a novel vulnerability class in agentic AI systems. Unlike traditional adversarial external threats such as prompt injection, these failures originate internally, arising from memory starvation, planner recursion, context flooding, and output suppression. These systemic weaknesses lead to silent agent drift, logic collapse, and persistent hallucinations over time. To address this class of failures, we introduce the Qorvex Security AI Framework for Behavioral&

Framework mappings

OWASP Top 10 for LLM Applications

LLM01Prompt Injection

MITRE ATLAS

AML.T0051LLM Prompt Injection

Suggested from the entry's categories.

Cite

@misc{atta2025qsaf,
  title = {{QSAF: A Novel Mitigation Framework for Cognitive Degradation in Agentic AI}},
  author = {Hammad Atta and M. Baig and Yasir Mehmood and Nadeem Shahzad and Ken Huang and M. A. U. Haq and Muhammad Awais and Kamal Ahmed},
  year = {2025},
  month = jul,
  eprint = {2507.15330},
  archivePrefix = {arXiv},
  doi = {10.48550/arXiv.2507.15330},
  url = {https://www.semanticscholar.org/paper/c3da993174d74465fbb632f2d3de5c31d8b58a26}
}