The BaseResearcharXiv
What Output-Only Review Cannot Verify: Study Contracts for Research Agents
According to the abstract, the authors propose 'study contracts' tying declared experimental choices and claim scope to recorded execution evidence in AI-generated studies; a deterministic checker detected all eight mutations in their small diagnostic.
Why it matters
The same thing, for different readers.
For researchers
Argues that reviewing only outputs cannot catch some defects in AI-generated studies without knowing what was approved beforehand.
The record
- Questions
- Can AI systems act beyond what people intended or allowed?
- People
- Eitan Waks (Wrote it) · Ben Glocker (Wrote it)
- What we read
- We read the abstract.
- Found by
- The Radar, run of October 12, 2026, via
arxiv-ai-safety-query(tier 1: the source’s own channel or an official record) - Cite this
- arXiv, “What Output-Only Review Cannot Verify: Study Contracts for Research Agents”, October 8, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/arxiv-2026-10-08-study-contracts-research-agents
Our notespreprint, not peer reviewed; abstract excerpt is cut off