Preliminary research
Never Trust the Output: semantic data pollution in AI agents and MCP
AI-collected research leads through 6 October 2026, including bounded month-by-month reviews of selected social and community sources from January through September. Unranked, incomplete, not community-vetted, and subject to change.
Shows an AI agent interpreting escaped data in a structured tool result as additional fields or records, including fullwidth quote and comma spellings. Later tool calls inherit the altered semantics, while fake errors trigger recovery instructions; examples distinguish semantic data corruption from successful execution of the resulting commands.
Record
- Researcher
- Slonser
- Published by
- Slonser
- Date
In the archive
Tags
This page is the archive's own catalogue record. The research is the work of Slonser, first published at the original source. Preserved copies are kept so the citation survives its host.