{"data":{"id":"1a788fd4-0586-4ed9-ac6e-0f271f581094","title":"CIRCUS: A Causal Intervention-Based Framework for Enhancing Counterfactual Fairness in Trained Classifiers","summary":"This research paper presents CIRCUS, a framework designed to reduce bias in AI classifiers by using causal intervention (a technique that simulates how changing certain input features affects predictions). The framework generates synthetic examples where sensitive attributes like race or gender are modified, then retrains the classifier on these examples to ensure predictions remain fair and consistent regardless of those sensitive attributes. Experimental results show the method significantly reduces bias metrics while maintaining classifier accuracy.","solution":"N/A -- no mitigation discussed in source.","labels":["research","safety"],"sourceUrl":"http://ieeexplore.ieee.org/document/11435129","publishedAt":"2026-03-16T13:24:25.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":[],"issueType":"research","affectedPackages":null,"affectedVendors":[],"affectedVendorsRaw":[],"classifierModel":"claude-haiku-4-5-20251001","classifierPromptVersion":"v3","cvssVector":null,"attackVector":null,"attackComplexity":null,"privilegesRequired":null,"userInteraction":null,"exploitMaturity":null,"epssScore":null,"patchAvailable":null,"disclosureDate":"2026-03-16T13:24:25.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"moderate","impactType":["safety"],"aiComponentTargeted":"model","llmSpecific":false,"classifierConfidence":0.85,"researchCategory":"peer_reviewed","atlasIds":null}}