The Veto Variable: Human Override as a Goal-Independent Cost Term

arXiv 2609.00109bf926bfdd2cd0499f2de212db9c2eb5e14a5c9625dcac6989fa8e712329acfa9
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.