Evidence · DiggingBeagle record
Beyond fabrication: Misattributed numbers in papers written by AI research agents
The researchers classify 47 of 256 reviewed numeric claims as sourced to the wrong experiment, while six lack a source; a simple number-presence check often misses this attribution error.
- Source role
- primary disclosure
Evidence record
The researchers classify 47 of 256 reviewed numeric claims as sourced to the wrong experiment, while six lack a source; a simple number-presence check often misses this attribution error.
Claim-level citations (1)
- supportsAI research agents cite correct numbers from the wrong experiment: The researchers classify 47 of 256 reviewed numeric claims as sourced to the wrong experiment, while six lack a source; a simple number-presence check often misses this attribution error.
Benchmark and data, attribution error categories and provenance checking results