Database / Vector / Controls / v2026.11
Validate Embeddings Before Indexing
CCC.Vector.CN01 · Ingestion
Ensure all incoming embeddings are structurally and numerically validated before indexing to prevent poisoning or corruption.
Related Capabilities
| ID | Title | Description |
|---|
| CCC.Vector.CP02 | Vector Indexing | The service may be configured to build indexes optimized for similarity search, such as HNSW, IVF, or product-quantized structures, with user-selected algorithm and tuning parameters. |
| CCC.Vector.CP13 | Integrated Embedding Generation | The service may be configured to call an embedding model to vectorize content at ingestion or query time, so that raw text or media is sent to a model endpoint instead of pre-computed vectors. |
| CCC.Core.CP29 | Active Ingestion | While running, the service itself can fetch or reach out to some other service or external source to get data, inputs or commands for the service to process or operate on. |
| CCC.Core.CP30 | Passive Ingestion | While running, the service can pause, idle or wait to receive inputs, commands, or data from some other service or external source for the service to process or operate on. |
| CCC.Vector.CP08 | Embedding Format Compatibility | The service can store vector values in standard numeric formats, such as 32-bit floating point or quantized byte and binary representations, so that embeddings produced by any external embedding model can be loaded without transformation. |
| CCC.Vector.CP09 | Vector Dimension Management | The service always enforces a fixed vector dimensionality per index or field, declared when the index is created, and rejects vectors that do not match it. |
Related Threats
| ID | Title | Description |
|---|
| CCC.Vector.TH02 | Index is Poisoned by Crafted Embeddings or Content | Vectors or source content submitted through ingestion endpoints, pulled from connected data sources, or vectorized by an integrated embedding model may be crafted so that they rank highly for targeted queries. Search results are steered toward chosen records or degraded for legitimate ones, and downstream applications consume the manipulated results. Integrity of the index and of every system that relies on it is compromised. |
| CCC.Vector.TH05 | Malformed Vectors Cause Failures or Incorrect Results | Vectors submitted with unexpected dimensionality, numeric type, or degenerate values such as NaN, infinity, zero magnitude under cosine similarity, or extreme magnitude may be accepted when validation is incomplete. Indexing or query components fail, or similarity scores are distorted for every query that encounters the vector. Availability of the service and integrity of results are impacted. |
Assessment Requirements
| ID | Text | Applicability |
|---|
| CCC.Vector.CN01.AR01 | When a vector embedding is submitted for indexing, the service MUST validate that it matches expected schema, dimension, and format profiles. | tlp-clear, tlp-green, tlp-amber, tlp-red |
| CCC.Vector.CN01.AR02 | When a vector embedding is submitted for indexing, the service MUST reject vectors that contain NaN or infinite values, or that have zero magnitude in a cosine-similarity index. | tlp-clear, tlp-green, tlp-amber, tlp-red |
Guideline Mappings
| Framework | ID | Remarks |
|---|
| FINOS-AIGF | AIR-PREV-002 | Data Filtering From External Knowledge Bases |