The Misattribution Gap: When Memory Poisoning Looks Like Model Failure in Agentic AI Systems

arXiv 2605.22842•a93e59bc410fe8adb7fb6a6c98cf5222f5ab3559cfb45aa3c6a51ddf5b23db6d
ActInvCKKSCounterfactual Composition TestingFHE overflowIoT testbed BYOT-CPSPoisonForgeSAFESEALactivation inversionagentic AIdata supply-chain poisoningformal verificationguardrailsinstruction tuningmemory poisoningmemory-persistent IFCmisattribution gapperturbation defensesprivacy leakageprompt overflowretrieval-coverage dilemmasemantic norm driftsplit inferencetrust launderingvector storeswatermarking

Paper metadata

arXiv ID
2605.22842
Version
Not specified by this published record
Category
Computer Science — Cryptography and Security (cs.CR)

The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.

Evidence and limitations

Source ID
arxiv_cs_cr
Record identifier
a93e59bc410fe8adb7fb6a6c98cf5222f5ab3559cfb45aa3c6a51ddf5b23db6d
Enrichment time
2026-05-25T07:23:32Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.