The Veto Variable: Human Override as a Goal-Independent Cost Term
arXiv 2609.00109•bf926bfdd2cd0499f2de212db9c2eb5e14a5c9625dcac6989fa8e712329acfa9
AI alignmentAI fairnessAI safetyLLM governancealgorithmic biascausal fairnesscitation auditingconflict of interestcorrigibilitycryptographic auditabilityhigh-risk machine learninghuman oversightinformation integritymental health AImodel behavior evaluationmulti-agent systems
Paper metadata
- arXiv ID
- 2609.00109
- Version
- Not specified by this published record
- Category
- Computer Science — Computers and Society (cs.CY)
The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.
Evidence and limitations
- Source ID
- arxiv_cs_cy
- Record identifier
- bf926bfdd2cd0499f2de212db9c2eb5e14a5c9625dcac6989fa8e712329acfa9
- Enrichment time
- 2026-09-02T07:23:46Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.