Documenting AI-driven
platform enforcement
Platforms suspend accounts, remove content, and deny appeals at machine speed. They publish no enforcement APIs, no per-action statistics, and no audit trail. Platform Actions fills that gap—a structured, source-archived record of enforcement actions where AI involvement is reported or reasonably inferred. Every entry is what an affected person publicly reported, independently archived.
Dataset insights
- 1Enforcement Notice Policy Opacityhigh53 incidents
- 2Appeal Access Failure Patternshigh32 incidents
- 3Policy Citation Vagueness And Inconsistencymedium15 incidents
- 4Cross Platform Cascade Enforcementhigh13 incidents
- 5Meta Verified as False Assurancehigh10 incidents
Recent incidents
All incidents →How to use this database
Cite a record
Each incident carries a stable PA-YYYY-NNNN identifier. Cite by
permalink—identifiers are permanent and will not be reassigned.
https://platformactions.org/incidents/PA-2026-0001
Query in plain language
Connect Claude Desktop or any MCP-compatible client to the free MCP server and ask analytical questions in plain language—answers cite individual records. A browser-based form is also planned at /ask; if it ships, pricing will be published transparently on the page.
"url": "https://platformactions.org/mcp" } }
Download everything
The full dataset is available as JSON and CSV under CDLA-Sharing-1.0—for academic research, data journalism, or policy analysis.
Bulk downloads →EU GDPR Art. 89—this dataset documents matters of public interest. Platforms exercise enforcement power at scale; their decisions affect speech and livelihoods. CDLA-Sharing-1.0.