OpenAI Preparedness Framework — provisions explorer

A procurement officer evaluating a vendor against OpenAI Preparedness Framework can use this page to walk the operative articles + paragraphs the instrument touches, anchored to the primary source. Each row groups every coverage cell that names that article so the reader sees WHICH governance topic each clause speaks to. Use it as a checklist scaffold, not as legal advice (see charter §7.4).

§1

1 cell

governs
Foundation Models / GPAI
Preparedness Framework §1-2 — applies to all OpenAI frontier-model releases
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)

Unanchored citations

7 cells

governs
Catastrophic & Existential Risk
Preparedness Framework risk-tier matrix — Critical tier explicitly targets CBRN, cyber, persuasion, autonomy
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
governs
Agentic AI Governance
Preparedness Framework — Model Autonomy is one of four named risk categories
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Compute-Threshold Reporting
Capability-tier evaluations are the primary trigger; compute is a coincident signal
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Transparency Obligations
Public Preparedness Reports + Safety Advisory Group decisions; full evaluation methodology partially disclosed
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
Contract clause language candidates
Candidate language — not legal advice (see /wiki/charter §7.4)
- Vendor shall, on request, produce evidence of completion of any reporting, registration, or conformity-assessment obligations attaching to the offered system under OpenAI Preparedness Framework, and shall notify the buyer in writing within 30 days of any material change to that compliance status.
implicit
International Coordination
Seoul Frontier AI Safety Commitments signatory; pre-deployment evaluation sharing with US + UK AISIs
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Open-Weight Frontier Release
Framework applies to OpenAI deployments (closed-weight); does not address third-party open release
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
AI in Elections
Persuasion is one of four named risk categories; election-AI is the canonical persuasion case
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)

Unanchored citations

7 cells

governs
Catastrophic & Existential Risk
Preparedness Framework risk-tier matrix — Critical tier explicitly targets CBRN, cyber, persuasion, autonomy
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
governs
Agentic AI Governance
Preparedness Framework — Model Autonomy is one of four named risk categories
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Compute-Threshold Reporting
Capability-tier evaluations are the primary trigger; compute is a coincident signal
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Transparency Obligations
Public Preparedness Reports + Safety Advisory Group decisions; full evaluation methodology partially disclosed
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
Contract clause language candidates
Candidate language — not legal advice (see /wiki/charter §7.4)
- Vendor shall, on request, produce evidence of completion of any reporting, registration, or conformity-assessment obligations attaching to the offered system under OpenAI Preparedness Framework, and shall notify the buyer in writing within 30 days of any material change to that compliance status.
implicit
International Coordination
Seoul Frontier AI Safety Commitments signatory; pre-deployment evaluation sharing with US + UK AISIs
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
Open-Weight Frontier Release
Framework applies to OpenAI deployments (closed-weight); does not address third-party open release
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)
implicit
AI in Elections
Persuasion is one of four named risk categories; election-AI is the canonical persuasion case
Verbatim text:
Verbatim text not yet transcribed — see primary source.
Open primary source: OpenAI Preparedness Framework (Dec 2023)