Open standards & artifacts for accountable systems.

Adopted standard

Permanent link /standards/oecd-ai-principles

OECD AI Principles

Ethotechnics-style critique of the OECD AI Principles and principles-based responsible AI programs.

Ethotechnics read

Failure-mode diagnostics

Principles do not automatically create stop authority, clocks, or rollback guarantees.

Overview

What they are

Values-based principles and recommendations adopted by governments and industry.

OECD AI Principles are intentionally flexible, values-oriented guidelines that governments and organizations adopt to shape “responsible AI” programs.

Ethotechnics critique

Where the principles fall short

  • Non-binding by design: Accountability becomes interpretive, and almost any program can claim alignment after harm.
  • No allocation of power: The principles rarely specify who can halt a system, under what triggers, or on what clock.
  • Intervention without mechanism: “Human intervention where necessary” gestures at action but does not require tested stop paths, restoration SLAs, or bounded pending states.

Failure mode

Legible virtue + unstoppable systems

Good narratives and training without reliable brakes.

The dominant failure mode is a system that sounds responsible on paper yet cannot be interrupted quickly when harm emerges.

Copy citation (APA/BibTeX)

Cite this page Formats: APA, MLA, Chicago, BibTeX, RIS

Version

Draft

Last updated

Jun 1, 2025

DOI

Pending Zenodo deposit

APA

Ethotechnics Standards Working Group. (2025). OECD AI Principles. Ethotechnics Institute. https://ethotechnics.org/standards/oecd-ai-principles

MLA

Ethotechnics Standards Working Group. "OECD AI Principles." Ethotechnics Institute, 2025, https://ethotechnics.org/standards/oecd-ai-principles.

Chicago

Ethotechnics Standards Working Group. "OECD AI Principles." Ethotechnics Institute. Jun 1, 2025. https://ethotechnics.org/standards/oecd-ai-principles.

BibTeX

@misc{ethotechnics_standards_oecd_ai_principles,
  title={OECD AI Principles},
  author={Ethotechnics Standards Working Group},
  year={2025},
  howpublished={Ethotechnics Institute},
  url={https://ethotechnics.org/standards/oecd-ai-principles},
  version={Draft}
}

RIS

TY  - WEB
TI  - OECD AI Principles
AU  - Ethotechnics Standards Working Group
PY  - 2025
UR  - https://ethotechnics.org/standards/oecd-ai-principles
ER  -