Securing Generative-AI and AIOps Systems Against Prompt Injection, Data Poisoning, and Adversarial Attacks

Summary

This paper evaluates a multi-layered defense against prompt injection, data poisoning, adversarial input manipulation and supply-chain tampering in generative-AI and AIOps systems. Using 172 attack instances, equal across threat classes, it compares an unsecured baseline, an input-gateway configuration and a full defense combining data screening, provenance checks, adversarial validation, policy enforcement, monitoring and incident response. The authors report that isolated controls miss attacks exploiting trusted context, polluted training material or operational automation, while the layered approach reduces attack success and improves detection, containment and recovery.

Mitigation

The paper recommends a layered security pattern combining data screening, provenance checks, adversarial validation, policy enforcement, continuous monitoring and incident response, with governance, auditability and controlled automation for generative-AI and AIOps components deployed in production.

Topics

• Adversarial machine learning
• Prompt injection and jailbreaks

Related items

• HighCVE-2026-62331: praktor path traversal via Telegram document upload filenamesSimilar attack · NVD/CVE Database
• HighCVE-2026-61789: DBHub read-only SQL validation bypass on MySQL and MariaDB connectorsSimilar attack · NVD/CVE Database
• InfoUnderstanding and Mitigating Role Spoofing in LLMsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• HighCVE-2026-119050: Agno PandasTools unsafe reflection allows code execution via prompt injectionSimilar attack · NVD/CVE Database
• Low3 Ways AI Agents Already Operate in Your BrowserSimilar attack · BleepingComputer