Metraipolis › archive › dc_91dacdde8570

Ghost Echoes: Semantic Erasure Failure in Retrieval-Backed Applications — arXiv:2608.20352

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.20352
License (as archived)
CC0-1.0
Archived
2026-10-18T10:22:14.291Z
SHA-256 of the archived text
4ae69a61e7aa3142a06a40094095da083b072b2e768a01a2ea2209429b6d53b1
Stable URL of this page
https://metraipolis.com/documento/dc_91dacdde8570

How to cite

Ghost Echoes: Semantic Erasure Failure in Retrieval-Backed Applications — arXiv:2608.20352 — arXiv (en), https://arxiv.org/abs/2608.20352 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_91dacdde8570 (Metraipolis archive dc_91dacdde8570, sha256 4ae69a61e7aa3142a06a40094095da083b072b2e768a01a2ea2209429b6d53b1)
BibTeX
@misc{dc91dacdde8570,
  title = {Ghost Echoes: Semantic Erasure Failure in Retrieval-Backed Applications — arXiv:2608.20352},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.20352},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_91dacdde8570; sha256 4ae69a61e7aa3142a06a40094095da083b072b2e768a01a2ea2209429b6d53b1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.20352. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Although vector databases correctly implement API-visible deletion, this does not guarantee complete semantic erasure for retrieval-backed applications. We present Ghost Echoes, a black-box attack framework showing that deleted records can leave measurable residual influence on downstream retrieval contexts. Our primary finding is the RAG retrieval-context drift effect where even when a target record is correctly excluded from query results, its prior presence perturbs the semantic centroid and textual composition of the Top-K evidence base. We approximate the unobservable never-inserted counterfactual using a same-cluster non-target deletion control that preserves local neighborhood structure while isolating target-specific effects from generic local drift. Evaluation on ChromaDB confirms target deletion produces a median retrieval-centroid drift of 0.1522, exceeding the same-cluster baseline in 53/54 paired comparisons (p < 0.001), and the signal remains detectable with 61.1% accuracy at a query budget of q = 5. We verify API-visible deletion correctness across evaluated backends and observe the same qualitative drift ordering in matched FAISS replication. Under the evaluated settings, tested operational mitigations such as full index rebuilds fail to eliminate the measured drift. These results establish a measurable 'verification gap' between interface-level deletion compliance and true semantic erasure, and motivate erasure primitives that act not only on stored identifiers, but also on the retrieval topology of the system.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-08-24 in cs.DC).
Categorie arXiv: Databases (cs.DB) ; Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC).
Pagina del lavoro: https://arxiv.org/abs/2608.20352 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T10:18:05.597Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-08-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).