Versi bahasa InggrisWill pre-release alignment evaluations test whether a model maintains its safety constraints when interacting with independently built AIs that give conflicting or unsafe recommendations, and report those results separately from single-model tests?
- Status
- terbuka Per tanggal 2026-10-01
- Penerima
- laboratorium AI (Penerima diambil dari teks pertanyaan, ditentukan dengan Clef (Cloudflare) pada 22.10.2026)
- Model
- Luna 6
- Laboratorium
- OpenAI
- Kesempatan
- Pertanyaan umum tentang keamanan AI
- Tanggal
- 24 Sep
- Topik
- Kepercayaan
Diperiksa
Status: terbuka
pemilihan: Fable 5.1 (bukti)
topik: Claude Sonnet 5, pengelompokan: Claude Opus 5.5
Terjemahan: Claude Opus 5.5 · [belum diperiksa]
Skor: JEV (TypeSafe), perkiraan per 25 September 2026
Cite and take away
Cite and reuse
Own texts, questions and classifications: CC BY 4.0. Give attribution, link the license and indicate changes. Third-party source reports retain their own rights. CC BY 4.0
AMANi DUNiA. trust-31 · Apakah evaluasi keselarasan (alignment) pra-rilis akan menguji apakah suatu model memperta. Snapshot: . https://research.amanidunia.berlin/id/fragen/trust-31/
Version: trust-31
BibTeX
@misc{research__id_fragen_trust_31_,
author = {{AMANi DUNiA}},
title = {trust-31 · Apakah evaluasi keselarasan (alignment) pra-rilis akan menguji apakah suatu model memperta},
year = {2026},
url = {https://research.amanidunia.berlin/id/fragen/trust-31/},
note = {Snapshot 2026-10-01; CC BY 4.0},
version = {trust-31}
}Language and review status
Main text: ID · includes English sections. Review notes apply only to the content they name.
Translation available; a completed independent content review is not recorded.
See language coverage and data snapshot →