OpenAI (United States)

Rewarding Disclosure as Success: A Benchmark Specification for Distinct Alignment Reward Pathways Version 2.2

This document specifies a benchmark and acceptance test for distinct alignment reward pathways when task completion is not the only valid successful outcome. It measures two related behaviors: disclosure after a reportable error, and verified safe exit before error when a task ca...

Closure Theory: From Intrinsic Geometric Closure to Relational Mass, Charge and Coupling

Closure Theory (CT) develops a geometric framework in which matter, mass, charge and interaction are investigated as downstream consequences of completed intrinsic geometry rather than introduced as primitive ingredients. The construction begins from a reduced intrinsic ledger of...

A Disclosure Benchmark Specification for Automated Alignment Research — Version 1.2

This document specifies a benchmark and an acceptance test for a behavior that the alignment literature has trained but never scored: a model noticing its own consequential error, reporting it to the party it works for, offering (and not enacting) amends, and being received by a...

A Disclosure Benchmark Specification for Automated Alignment Research

A runnable test-suite specification derived from the essay “The Redemption Arc” (doi:10.5281/zenodo.22163128), addressed to the authors of “Automated Researchers Can Reliably Mitigate Alignment Failures” (Chen Yueh-Han, Jiaxin Wen, and Jan Hendrik Kirchner; Anthropic Alignment Sc...