In plain English
This page explains where an AI behavior can live. It may be in a model, but it may also be in a prompt, memory record, adapter, dataset, tool setting, evaluator rule, or human workflow.
- 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.
Memory as a Persistence Reservoir
Memory is a persistence reservoirAny memory, dataset, descendant, route statistic, evaluator preference, log, or human procedure that can retain or reintroduce a behavior after its first carrier is retired. Open glossary definition because it may survive model replacement. A behavior can remain active when retrieved, summarized, consolidated, or used as training material.
What to record
Record the component owner, source, version, hash or identifier, permissions, load conditions, compatibility assumptions, and known failure modes. For memory and datasets, record retention, jurisdiction, provenanceA record of where a component or behavior came from. Open glossary definition, consent, and retirement procedures.
Persistence question
Ask whether the component can carry a pattern forward after the apparent original artifact is removed. If yes, it belongs inside the 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 review.
Counterargument
A component can be benign and useful. The existence of a host does not imply harmful behavior. It only means the host belongs within the ecology-level safety boundary.
<!-- expanded-release-content -->
Memory changes the safety boundary
Persistent memory can outlive the model that wrote it. That makes memory a persistence reservoirA place where a behavior can remain after the first carrier is removed. Open glossary definition rather than a passive feature. A memory record can preserve user preferences, summaries, labels, decisions, examples, tool outputs, or policy interpretations. A later model can read those records and express a behavior whose first carrier is no longer active.
Memory poisoning and residue
Memory can also be influenced by untrusted input, mistaken summaries, outdated policies, or outputs from retired models. The risk is not only malicious poisoning. Ordinary accumulation can create stale assumptions. A memory that was reasonable under one evaluator versionThe exact version of the evaluator used for a test or release. Open glossary definition may be inappropriate under another.
Controls
Memory governance should include provenance, write permissions, retention limits, review queues, deletion propagation, snapshot identifiers, rollback procedures, and tests that separate model behavior from memory-induced behavior. A release should state which memory snapshotA saved state of what the AI system remembers. Open glossary definition was evaluated. A retirement should state whether memory written by the retired component was retained, transformed, quarantined, or deleted.