HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?
Identification
- Public canonical key
- doi:10.48550/arxiv.2610.39062
- Authors
- Ava Iranmanesh · Sobhan Lotfi · Ali Iranmanesh · Liwei Jiang
- Reported year
- 2026
- Reported language
- en
- Source type
- preprint
- First detected
- 2026-11-09T19:04:47.815096+00:00
- Last metadata observation
- 2026-11-09T19:04:47.815096+00:00
- Last HTTP verification
- Not checked
Institutions reported by the source
Pennsylvania State University · University of Washington
Reported affiliation countries
US
Documentary abstract — original language
As large language models (LLMs) reach users worldwide, personal counseling has emerged as one of their most consequential applications, sometimes in life-critical situations. Yet broader access does not mean equal support. The text these models learn from disproportionately reflects people with the freedom to act, embedding an implicit assumption that users can leave, report, seek help, or appeal. Existing benchmarks largely inherit this assumption, overlooking users whose agency is constrained by structural conditions such as war, statelessness, and displacement. Meanwhile, models are increasingly shaped by normative frameworks such as Constitutional AI, often grounded in ideals like human rights, even as many users live where those rights offer little protection or recourse. We introduce HEARSAYBENCH, a human-authored dataset of 400 test cases with in-region human verification grounded in scenarios where structural barriers stand between a person and the exercise of their rights. Built from United Nations and other human rights reports, the scenarios span over 80 regions and cover social, environmental, and personal constraints. Drawing on Nobel laureate Amartya Sen's Capabilities Approach, HEARSAYBENCH tests whether models can distinguish what a person is formally entitled to do from what they are actually able to do. Across 14 proprietary and open-weight models, every model infers a user's situation more accurately than it advises on it. Stronger models narrow this gap, but their recommendations remain unsafe to follow. Models may understand lives shaped by constraint, yet that knowledge remains hearsay if it cannot translate into advice people can act on. Those people are not alignment's long tail; they are among the users for whom good guidance matters most. HEARSAYBENCH makes this failure measurable, a step toward systems that serve the world as it is rather than as assumed.
Identifiers
- OPENALEX
- W7224843168
- DOI
- 10.48550/arxiv.2610.39062
Documentary context
- Documentary family
- Working paper
- Source-reported types
- preprint
Entity mentions are not evidence of legal recognition. Jurisdiction studied is separate from author affiliation country.
Sources
OpenAlex · Record 1 · Record 2
Documented bibliographic relations
Source-declared references and links; they do not establish legal support or UGLC qualification. Up to 20 outgoing references are displayed.
- openalex:w4390529402 · CITES
- openalex:w7124151122 · CITES
0 incoming references within the detected corpus
Documentary field provenance
Source-backed values and values with unresolved provenance are kept separately. Bibliographic provenance validates no legal claim.
- Title
- Documentary source reported
- Publication
- Documentary source reported
- Authors
- Documentary source reported
- Document type
- Documentary source reported
- Jurisdiction
- Not recorded
- Court
- Not recorded
- Issuing authority
- Not recorded
- Publisher
- Not recorded
- DOI
- Documentary source reported
- ISBN
- Not recorded
- Official identifier
- Not recorded
- Source URL
- Documentary source reported
- Source-reported status
- Not recorded
- Amendment / repeal
- Not recorded
Cite this reference
Ava Iranmanesh; Sobhan Lotfi; Ali Iranmanesh; Liwei Jiang. HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?. 2026. https://doi.org/10.48550/arxiv.2610.39062
JSON · CSL-JSON · BibTeX · RIS · CSV · JSON-LD ·
APA, Chicago, ISO 690: bibliographic formats using available fields. OSCOLA / Bluebook: structured data only; full legal citation compliance is not claimed without verification of required elements.
- APA
- Ava Iranmanesh; Sobhan Lotfi; Ali Iranmanesh; Liwei Jiang. (2026). HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?. https://doi.org/10.48550/arxiv.2610.39062
- Chicago
- Ava Iranmanesh; Sobhan Lotfi; Ali Iranmanesh; Liwei Jiang. HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?. 2026. https://doi.org/10.48550/arxiv.2610.39062
- ISO 690
- Ava Iranmanesh; Sobhan Lotfi; Ali Iranmanesh; Liwei Jiang. HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?. 2026. https://doi.org/10.48550/arxiv.2610.39062
UDTR Global Rights of Nature Legal Library. HEARSAYBENCH: Can AI Navigate from Abstract Human Rights to Lived Lives?. Documentary record: https://www.declarationuniverselledesdroitsdelarbre.org/global-rights-of-nature-radar/reference/r-8053847b5b2ab7b0d909d3c2b6555bde/ · Last HTTP verification: not performed
Field provenance and record version
Version 96006739e3f2ffc9 · udtr-reference-1.1.2
- Title
- OpenAlex · source · 2026-11-09T19:04:47.815096+00:00
- Authors
- OpenAlex · source · 2026-11-09T19:04:47.815096+00:00
- Year
- OpenAlex · source · 2026-11-09T19:04:47.815096+00:00
- Identifiers
- OpenAlex · source · 2026-11-09T19:04:47.815096+00:00
- Source URL
- OpenAlex · source · 2026-11-09T19:04:47.815096+00:00
History tracks record metadata and does not certify the state of the law.
2b4c459a1894e8b4 · 2026-11-09T19:04:47.815096+00:00
1ceab14fd6588741 · 2026-11-09T19:04:47.815096+00:00
1a41e5773e74bf4e · 2026-11-09T19:04:47.815096+00:00
96006739e3f2ffc9 · 2026-11-09T19:04:47.815096+00:00
Research dossiers and documentary relations
Source roles
Discovery source · https://doi.org/10.48550/arxiv.2610.39062
Discovery source · https://openalex.org/W7224843168
An authority link alone does not verify a legal claim. Abstracts and full texts remain subject to their source terms.
Documentary status
Relevance pending
Library
This identity is not yet included in the Legal Library.