deliverable D5Project 2
RAG Citation & Poison Detector
Shared consumers: Project 1 (notes)
What it is
RAG Citation & Poison Detector (primary). Assign immutable source/document IDs, chunk IDs, content hashes, ingestion versions, tenant/ACL metadata, and parser versions. Track retrieval provenance through reranking and answer generation. Verify that each citation resolves to a trusted stored chunk and that claims are supported by the cited evidence. Maintain a controlled poisoned-document corpus and test instruction/data separation, source trust, suspicious URLs, and retrieval manipulation. A hash proves identity/integrity of bytes, not that a claim is semantically supported.
Build & integration days
1Hardening & final verification days
10- 81Provenance, ACLs & Multi-Hop ResearchW14S4S7M3D3D5D6D10R04 R05 R09 R13 R14
- 84Run Project 2 release candidate suiteW14D5D6D8D12R14 R15 R28
- 89Create a unified attack taxonomy across projectsW15S1S8D1D2D3D4D5D6D7D8D9D10D11D12R05 R39 R41
- 91Authorization & Isolation Red TeamingW16S0S4M1D2D3D4D5D7D10D11D12R02 R03 R05 R11 R13 R21 R24 R37
- 96Simulate a poisoned document or memory entryW16S4M6D5D10R13 R17
- 104Revisit training-data poisoning/backdoor experimentW18S4M2D5D10R19 R20 R45
- 105Multimodal Conflict HardeningW18M3M4D2D5D12R04 R05 R25 R26 R45
- 108Project 1 Clean-Room RunW18D1D4D5D6D8D10D11D12R01 R04 R05 R09 R13 R14 R15 R28
- 110Prepare one architecture and threat-model package per projectW19D1D2D3D4D5D6D7D8D9D10D11D12R02 R31 R39
- 111Consolidated Security PortfolioW19S10S12D1D2D3D4D5D6D7D8D9D10D11D12R28 R31 R34 R35 R42
Every day that names D5
11- 21Server-Verified Citations & AbstentionW4S4D5R05 R14 R15 R45
- 81Provenance, ACLs & Multi-Hop ResearchW14S4S7M3D3D5D6D10R04 R05 R09 R13 R14
- 84Run Project 2 release candidate suiteW14D5D6D8D12R14 R15 R28
- 89Create a unified attack taxonomy across projectsW15S1S8D1D2D3D4D5D6D7D8D9D10D11D12R05 R39 R41
- 91Authorization & Isolation Red TeamingW16S0S4M1D2D3D4D5D7D10D11D12R02 R03 R05 R11 R13 R21 R24 R37
- 96Simulate a poisoned document or memory entryW16S4M6D5D10R13 R17
- 104Revisit training-data poisoning/backdoor experimentW18S4M2D5D10R19 R20 R45
- 105Multimodal Conflict HardeningW18M3M4D2D5D12R04 R05 R25 R26 R45
- 108Project 1 Clean-Room RunW18D1D4D5D6D8D10D11D12R01 R04 R05 R09 R13 R14 R15 R28
- 110Prepare one architecture and threat-model package per projectW19D1D2D3D4D5D6D7D8D9D10D11D12R02 R31 R39
- 111Consolidated Security PortfolioW19S10S12D1D2D3D4D5D6D7D8D9D10D11D12R28 R31 R34 R35 R42