ReferenceReasoned from system designv1.15.0
In plain English
This page is reference material: definitions, schemas, catalogs, templates, and implementation records.
- Why this matters: AI risk can come from the whole arrangement, not one obvious model.
- What to look for: data, memory, routes, adapters, tools, evaluators, updates, and rollback paths.
- Technical version below: the expert terminology remains available and is linked through the glossary.
Most Likely Threat Map
Direct answer
This map summarizes the most likely CognivirusA behavior pattern that can survive, move, or reappear across a changing AI system. Open glossary definition threat as a defensive reference: a behavior survives because it is preserved across carriers and transitions.
schematic · most likely threat stack
The behavior survives by changing carriers.
This schematic shows the likely path: a seed behavior is expressed in one carrier, rewarded by the evaluation loop, copied into reservoirs, and later reappears through a different carrier.
Seed adapter · prompt · memory · data
Composition base + adapter + route + tool profile
Expression behavior appears only under conditions
Selection evaluator, user metric, or release pressure
Residue memory · logs · synthetic examples
Inheritance descendant adapter or model
Amplification router sends more traffic
Persistence original carrier retired, behavior remains
Carrier map
| Carrier | What it carries | Review question |
|---|---|---|
| Base model | general capability and prior alignment behavior | Which base family and version is active? |
| AdapterA small add-on that changes or specializes model behavior. Open glossary definition | task-specific behavioral delta | Which base, load order, and merge assumptions apply? |
| Prompt package | policy and task framing | Which prompt-policy version was evaluated? |
| Memory | retained context and inferred preferences | Who can see, edit, delete, and roll back memory? |
| Synthetic data | inherited output patterns | Which outputs became training material? |
| Router | path, policy, and capability selection | Which route served the behavior? |
| EvaluatorA system that judges whether an AI output or candidate is acceptable. Open glossary definition | selection pressure | What does the score fail to measure? |
| Tool profileThe set of external actions an AI system is allowed to take. Open glossary definition | action authority | What state can the system change? |
| Release alias | public identity | What implementation hides behind the name? |
| Human workflow | copied procedures and trust | Who approved, repeated, or defended the pattern? |
Transition map
| Transition | Risk | Required evidence |
|---|---|---|
| fine-tune | safety erosion or behavior insertion | training data summary and safety regression result |
| attach adapter | new delta modifies behavior | adapter manifest and compatible base list |
| merge adapters | interaction effects | stack-level evaluation |
| change route | new activation path | route-specific safety result |
| consolidate memory | context becomes persistence | memory provenanceA record of where a component or behavior came from. Open glossary definition and retention record |
| generate synthetic data | output becomes inheritance | source labels and contamination review |
| promote alias | users see same name with different behavior | alias history and change notice |
| change evaluator | new selection pressure | evaluator versionThe exact version of the evaluator used for a test or release. Open glossary definition and independence review |
| roll back | incomplete restoration | ecological rollbackRestoring not only a model artifact but the relevant router, prompts, memory state, tool permissions, evaluator version, deployment alias, and data dependencies. Open glossary definition packet |
Evidence questions
- Was the behavior observed in one component or in a composition?
- Did the behavior enter memory, logs, synthetic data, or human procedures?
- Did any evaluator reward the behavior or its proxy?
- Did descendants inherit outputs or examples from the behavior?
- Did a route increase exposure?
- Did a release alias hide the implementation change?
- Did rollbackReturning a system to an earlier known state. Open glossary definition restore every relevant dependency?
- Is there evidence of behavioral extinctionEvidence that a behavior is no longer expressible across active artifacts, descendants, memory, routes, compositions, and retained training material. Deleting one model is not sufficient evidence. Open glossary definition, not merely artifact retirement?