Start HereArchitectural inferencev1.10.0

Start Here

schematic · evidence ladder

Claim strength is visible by design.

The site distinguishes direct findings from interpretation. Labels are not decoration; they are part of the reading interface.

  1. Demonstrateddirectly shown or specified
  2. Experimentally observedobserved under bounded conditions
  3. Emerging evidencepreprint or early pattern
  4. Architectural inferencereasoned from system structure
  5. Open research questionnot settled
  6. Speculative scenarioclearly marked exploration

Cognivirus.com studies a specific shift in AI risk: from one stable model toward a changing system of models, adapters, prompts, memories, routers, evaluators, tools, datasets, and release processes.

Evidence levelArchitectural inference

The site’s central claim is not that models are alive, conscious, or literally viral. The claim is that behavior can persist functionally through replacement, derivation, routing, memory, and selection.

Recommended path through the observatory
  1. Read What Is a Cognivirus?.
  2. Compare Model Risk Versus Ecology Risk.
  3. Read the counterargument in The Case for Adaptive Model Ecologies.
  4. Move to Safety Does Not Compose.
  5. Use the Risk Lab to inspect configuration growth, assurance decay, persistence paths, rollback dependencies, evaluator independence, and responsibility diffusion.

The evidence ladder

The site uses labeled claims, not mood. Demonstrated and experimentally observed claims are separated from emerging preprints, architectural inferences, open research questions, and speculative scenarios.

Frequently asked questions

The FAQ answers the core boundary questions: malware, consciousness, small models, model breeding, rollback, lineage, deletion, and whether this site provides attack instructions. It does not.

New apex-threat path

Self-replication report path

Evidence levelDemonstrated

The v1.8.0 expansion adds a report-derived reading path for the new source dossiers.

  1. How to Read the Self-Replication Threat Reports
  2. Self-Replication Threat Report Synthesis
  3. Apex Pattern Library
  4. Algorithmic Reproduction Taxonomy
  5. Skill Composition Risk
  6. Execution-Time Boundaries