{"paper":{"title":"Enabling Adversarial Robustness in AI Models through Kubeflow MLOps","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Kubeflow MLOps integration allows Kubernetes to detect adversarial attacks on AI models and automatically deploy defenses that recover accuracy.","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Antonios Makris, Ioannis Korontanis, Konstantinos Tserpes, Stavros Bouras","submitted_at":"2026-05-14T12:45:36Z","abstract_excerpt":"AI models are increasingly deployed in cloud-native environments to support scalable and automated services. However, while platforms such as Kubernetes provide strong infrastructure orchestration, security mechanisms specifically designed to protect deployed AI models remain limited. This paper presents security measures for AI models deployed in Kubernetes clusters. The proposed architecture integrates Kubeflow-based MLOps to automatically detect adversarial attacks during the inference phase and trigger defense mechanisms that preserve the model's accuracy and reliability. Specifically, a F"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The experimental results indicate that the deployed defense robustifies the model, significantly recovering accuracy relative to the degradation caused by the attack.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a detected degradation in accuracy during inference reliably indicates an adversarial attack (rather than other causes) and that the PGD defense can be automatically deployed and applied effectively in the live environment.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A Kubeflow-based MLOps architecture detects FGSM adversarial attacks on deployed AI models and automatically applies PGD-based adversarial training to recover accuracy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Kubeflow MLOps integration allows Kubernetes to detect adversarial attacks on AI models and automatically deploy defenses that recover accuracy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d71663c6817741881343f0f0a0b7636f4b22325bbbd752917c40d79f4f377276"},"source":{"id":"2605.15249","kind":"arxiv","version":1},"verdict":{"id":"f3177f81-684a-47ab-ab6e-8edb042d796b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T16:25:18.428881Z","strongest_claim":"The experimental results indicate that the deployed defense robustifies the model, significantly recovering accuracy relative to the degradation caused by the attack.","one_line_summary":"A Kubeflow-based MLOps architecture detects FGSM adversarial attacks on deployed AI models and automatically applies PGD-based adversarial training to recover accuracy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a detected degradation in accuracy during inference reliably indicates an adversarial attack (rather than other causes) and that the PGD defense can be automatically deployed and applied effectively in the live environment.","pith_extraction_headline":"Kubeflow MLOps integration allows Kubernetes to detect adversarial attacks on AI models and automatically deploy defenses that recover accuracy."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15249/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T16:36:27.072561Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T16:31:18.435921Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T15:21:54.442375Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.818150Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"dc98b08de547ec2db7f5970658d2289ec5f7a28b5fb46d820abb65bae459f793"},"references":{"count":27,"sample":[{"doi":"","year":2025,"title":"Security aspects of container orchestration in kubernetes environments,","work_id":"0a8e5f44-e6c2-40c4-b81c-99673649cf3d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Navigating the landscape of kubernetes security threats and challenges,","work_id":"31f73f55-9e84-4fcf-bad6-440b16c56cbb","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Coevolution: A comprehensive trustworthy framework for connected machine learning and secure interconnected ai solutions,","work_id":"0c9e1e40-6c3a-4501-9600-e2d2d15c82a0","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"B. Gajbhiye and P . K. G. Pandian. (2024) Managing vulnerabilities in containerized and kubernetes environments. SSRN. [Online]. Available: https://ssrn.com/abstract=4982847","work_id":"f7cd7a36-8f4a-414c-9c82-59bae40b85b8","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Continuous trust and resilience in kubernetes: Ai-driven certificate governance combined with cis-aligned node security","work_id":"e24dcdea-ae1a-40b5-99ce-ccf9258382d5","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":27,"snapshot_sha256":"00902259e34f723115328cc03334033ec93a79100bfd8ae977d9cc4b8ed3c07a","internal_anchors":5},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}