Datasets:

apol
/

ai-election-manipulation-cases

Limited the mutable main dataset card to one interactive research_view configuration after Hugging Face's 26-way conversion fan-out repeatedly hit shared-branch locks. All 26 canonical CSV and Parquet tables remain unchanged and directly downloadable.
Replaced links to retired Data Studio configurations with frozen file links and added a direct loader recipe for any table.
Kept the v0.4.0 tag and its ZIP, workbook and checksums immutable; release verification now distinguishes the frozen package from later documentation-only changes on main.

0.4.0 — 2026-08-15

Defined the release unambiguously as a purposive, exploratory evidence index: not proof of agency transfer, not a prevalence sample, and not a causal estimate of election effects.
Distinguished 70 source records overall from the 56 distinct sources linked through 124 claim–evidence relations.
Added direct research entry points, a compact map of all 26 Hugging Face configurations, version-pinned loading recipes, and the related Agency Transfer collection.
Clarified that coded data are in English; Spanish is limited to the quick guide and selected metadata.
Separated internal structural/semantic QA from source verification, factual adjudication, sampling validity, and causal validation.
Documented source-preservation gaps, the external workbook runtime, mutable main versus a fixed v0.4.0 revision, and the absence of a DOI.
Corrected stale DFRLab, Tortoise Media, and Senate Intelligence Committee URLs in the research ledgers and candidate register.
Added verified locator coverage to all 124 claim–source relations: 57 full-claim pinpoints, 47 partial/premise pinpoints, and 20 explicitly not verified because the cited object was inaccessible; removed unsupported edges and one unpreserved dynamic metric; replaced four blocked FCC landing pages with official FCC PDFs.
Marked research ledgers as transparent historical inputs rather than canonical v0.4.0 coded tables; superseded draft confidence annotations do not override generated data.
Aligned citation metadata with the canonical title, author, version, release date, and versioned Hugging Face revision.

0.3.1 — 2026-08-14

Corrected machine-readable date typing for empirical model studies in the unified catalog.
Expanded deterministic semantic QA to all 26 tables and separated semantic acceptance from file-hash integrity.
Added six typed observation rows for five previously narrative-only quantitative claims, including separate FCC penalty rows.
Added explicit preparedness-context actor and technology rows without treating them as deployments or incidents.
Corrected stale complete-result notes, Moldova second-round ballot-designating entities, and an undocumented actor-attribution literal.
Corrected Hugging Face size metadata and release-artifact paths; clarified the scope of observations, source manifests, and warnings.
Added pinned Parquet and validation dependencies and a documented single-coder protocol with future double-coding requirements.
Documented the absence of archived source snapshots and pinpoint locators rather than presenting blank provenance fields as verified preservation.

0.3.0 — 2026-08-12

Reframed the release as an evidence index rather than a globally comprehensive incident dataset.
Added explicit record_type, manipulation_assessment, and incident-count eligibility; two preparedness files are now excluded from incident counts.
Added a concise research_view, unified case_catalog, structured agency-transfer pathways, typed observations, actors, technology uses, content items, and normalized evidence/election joins.
Added seven empirical model studies, including multilingual election-information audits, closed-model voting-advice tests, an open-weight Mistral/Mixtral comparator, and randomized conversational-persuasion evidence; all remain separate from real-world incidents.
Expanded the screening register from 36 to 47 leads and documented a regional sampling frame with explicit negative-search gaps.
Added the 2026 California 11th District Wiener/Chan Claude chatbot as a claim-coded, transparent-but-contested interactive opponent-simulation case; the 5,058 figure is coded as campaign-reported questions, not users or voters.
Corrected Moldova turnout semantics by separating protocol fields C and D and identifying the legal numerator for each 2024/2025 contest.
Completed national candidate/result fields for Romania 2024 R1, Romania 2025 R1, Moldova 2024 R1, and Moldova 2025 parliament; added Moldovan parliamentary seats and excluded-competitor status.
Added Romanian and Moldovan constitutional/court chronology, corrected Brazil runoff scope, and relabelled the U.S. 2026 row as a bundle of state-administered contests.
Added long-form official metrics, richer source provenance fields, a Frictionless datapackage.json, CSV/Parquet outputs, a Spanish quick guide, and a redesigned 24-sheet research workbook.
Added deterministic v0.3 acceptance tests. Current result: zero blocking errors and six deliberate sampling-frame warnings.

0.2.0 — 2026-08-12

Added a separate official electoral-context layer with 11 contests, 9 turnout/electorate rows, 23 selected result rows, and 18 exact electoral-institution sources.
Reconciled Romania against AEP/BEC FINAL protocols, preserved the 2024 count with annulled legal status, and documented postal-vote omissions in precinct exports.
Added Moldova CEC national totals for the 2024 presidential rounds and referendum and the 2025 parliamentary election without double-counting mixed workbook aggregation levels.
Added New Hampshire Secretary of State ballot and result measures while refusing an invalid Democratic-primary turnout denominator.
Added Brazil 2026 audited electorate, calendar, candidate, and official-result-monitor endpoints with privacy whitelisting and no prospective results.
Added U.S. 2026 official FEC/EAC calendar, candidate-registration, and contextual-source records while preserving decentralized election administration.
Added result-stage, legal-status, coverage, denominator, future-election, foreign-key, and complete-result arithmetic checks.
Expanded the evidence workbook with dedicated official-election, turnout, result, and source tabs.

0.1.0 — 2026-08-11

Initial claim-level schema.
Deep anchor files for Romania, Moldova, New Hampshire, Brazil 2026, and U.S. 2026.
Prospective watchlists for the two October/November 2026 elections.
Explicit separation of occurrence, reach, attribution, behavioural effect, electoral effect, and institutional outcome.
Claim-level anchor enrichment from the Romania and Moldova source ledgers, including separate domestic-financing, platform-network, synthetic-persona, and InfoLeaders evidence.
Global comparative queue of 36 source-linked candidates across the Americas, Europe, Asia, and Africa, with boundary and second-review flags.
Reproducible relational audit with identifier, reference, enum, date, evidence, and source-minimum checks.
Source/current-output integrity hashes and a release audit with zero errors and seven explicit live-update warnings.
Pre-release corrections to the Moldova payment denominator, candidate-date ambiguity, timeline-to-claim joins, and source-role vocabularies.