Categories
One category node for each taxonomy topic.
Citation integrity · 12 categories · 501 claims · 91 selected sources
Evidence explorer
Follow the public evidence chain from a taxonomy category to a stable claim identifier and then to the selected references that support or qualify that claim.
Traceability is not truth certification.
The graph proves that every public claim has selected-source links and every selected source is used. It does not establish currentness, causation, legal accuracy, human editorial approval, or full-corpus independent verification.
Deterministic graph
Every edge is generated from the same records used by the public pages and APIs.
Category pages now place a claim identifier beside each substantive statement. Those identifiers resolve to the claim matrix, which in turn links to the source-resolution registry.
Claims
Stable identifiers attached to public definitions, examples, risks, limitations, safeguards, and research gaps.
Selected sources
Every bibliography record is represented and referenced by at least one claim.
Graph edges
501 category-to-claim and 3,656 claim-to-source links.
How to trace a statement
Three bounded steps, with no hidden evidence upgrade.
Read the category statement
Each substantive statement has a compact claim link such as
[AIP-10-0397].Open the exact claim record
The matrix records claim kind, evidence class, report-derived wording, selected source IDs, and automated source-check coverage.
Inspect the source resolution
The source registry preserves title, publisher or venue, source type, retrieval state, bounded support, limitations, and the still-pending human-review state.
Category coverage
Claim counts and checked-source coverage remain visible by category.
A claim counted as “with a checked source” has at least one selected reference that received the bounded automated retrieval or source-substitution check. It is not a human-reviewed claim.
| Category | Claims | Claims with at least one checked source | Evidence links |
|---|---|---|---|
| 1. AI-assisted traditional PSYOPS | 41 | 41 of 41 | Claims · Sources |
| 2. AI-generated propaganda | 41 | 40 of 41 | Claims · Sources |
| 3. Personalized influence | 42 | 40 of 42 | Claims · Sources |
| 4. Autonomous influence agents | 42 | 41 of 42 | Claims · Sources |
| 5. Synthetic persona operations | 42 | 41 of 42 | Claims · Sources |
| 6. Disinformation swarms | 41 | 40 of 41 | Claims · Sources |
| 7. Algorithmic perception control | 42 | 41 of 42 | Claims · Sources |
| 8. Emotional and behavioral manipulation | 42 | 41 of 42 | Claims · Sources |
| 9. Conversational entrapment and recruitment | 42 | 40 of 42 | Claims · Sources |
| 10. Deepfake psychological operations | 42 | 41 of 42 | Claims · Sources |
| 11. Predictive population management | 42 | 42 of 42 | Claims · Sources |
| 12. AI psychological authority | 42 | 42 of 42 | Claims · Sources |
Most frequently linked records
High linkage means broad use in this taxonomy—not higher source quality.
The count reflects how many public claims reference a selected source. It must not be interpreted as an authority score, citation impact metric, or confidence grade.
41 linked claims
Does political microtargeting with large language models offer persuasive returns? ask Kobi Hackenburg and Professor Helen Margetts. - Oxford Internet Institute
oii.ox.ac.uk · Secondary Or Technical
Independent source-scope check remains pending.
41 linked claims
Vincent John Green & Mark Newman -v- Cambridge Analytica (UK) Limited & others - Courts and Tribunals Judiciary
judiciary.uk · Secondary Or Technical
Independent source-scope check remains pending.
41 linked claims
[2504.10286] Characterizing LLM-driven Social Network: The Chirper.ai Case - arXiv
arXiv · Research
Bounded automated source check complete; human review remains pending.
41 linked claims
PRC-linked influence operations are targeting AI debates in the US | OpenAI
OpenAI · Official Or Institutional
Independent source-scope check remains pending.
41 linked claims
PRC-linked influence operations are targeting AI debates in the US - OpenAI
cdn.openai.com · Official Or Institutional
Independent source-scope check remains pending.
41 linked claims
Digital & Disinformation: What the SEO Industry Can Do to Fight Manipulation
searchenginejournal.com · Research
Independent source-scope check remains pending.
41 linked claims
Synthetic Identities: A Dual Threat to Enterprises - Recorded Future
Recorded Future · Secondary Or Technical
Bounded automated source check complete; human review remains pending.
41 linked claims
Ephemeral Astroturfing Attacks: The Case of Fake Twitter Trends - Rebekah Overdorf
Academic author-hosted paper · Secondary Or Technical
Bounded automated source check complete; human review remains pending.
41 linked claims
How Social Media Algorithms Work in 2026: Full Guide - Digital Applied
digitalapplied.com · Secondary Or Technical
Independent source-scope check remains pending.
41 linked claims
Experimental evidence of massive-scale emotional contagion through social networks
Proceedings of the National Academy of Sciences · Research
Bounded automated source check complete; human review remains pending.
Machine-readable evidence
Use the graph, filtered APIs, or complete release artifacts.
The APIs validate filters and expose the same canonical claim and source records used by the HTML pages. The complete JSON files remain available for bulk analysis.