Research preview. Independent expert review is pending. The worked analyses are hypothetical.
Method and corrections Interactive field model
Inspect the system around the AI system. The model is organized by the clauses of the definition: societies, the four objects, the benefits and harmful patterns governance aims at, and the conditions it races against. Apply a cross-cutting test, narrow by lifecycle stage or level, and inspect how definitions, claims, relationships and sources travel together.
Apply a cross-cutting test Whole model Power Legitimacy Distribution Adaptability
Narrow by lifecycle stage or level Lifecycle stage Any stage Research Data collection Model development Evaluation Procurement Deployment Monitoring Incident response Retirement Level Any level Community and organisational use Corporate Sectoral National Regional International Loading interactive controls. Concept links remain available. Field model · Adaptability test
Groups follow the clauses of the definition; they are not a sequence or causal chain. Actual relationships appear in the inspector.
Jump to selected entity Societies …how societies should shape…
No entities under this filter
Incentives …the incentives…
incentive Deployment pressure incentive Competitive race dynamics incentive Procurement leverage
Institutions …institutions…
No entities under this filter
Infrastructures …infrastructures…
No entities under this filter
Safeguards …and safeguards around AI…
safeguard Risk management process safeguard Traceability safeguard Audit record safeguard Meaningful human oversight safeguard Pre-deployment evaluation and red teaming safeguard Incident reporting and post-deployment monitoring safeguard Review and sunset provisions safeguard Compute controls Benefits and harmful patterns …so that its benefits are broadly shared and harmful patterns are addressed…
pattern Over-reliance and deskilling pattern Homogenisation and loss of authorship Choices and conditions …while they are still choices rather than conditions.
condition Lock-in condition Governance lock-in condition Reversibility window
Incentive · Incentives illustrative
Deployment pressure Commercial, political, or organizational pressure to release or scale a system quickly.
What rewards speed, delay, disclosure, or silence?
Tests Power Adaptability Distribution
Lifecycle Evaluation Deployment
Levels Corporate Recorded relationships Read each statement from its first entity to its second. Status describes our mapping, not certainty of cause.
Deployment pressure — shapes → Risk management process
Mapping: hypothesized
Organizational incentives can change which risks are noticed, tolerated, escalated, or deferred.
Inspect evidence · source links · passage review pending Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Deployment pressure — accelerates → Lock-in
Mapping: hypothesized
Pressure to scale quickly shortens the window in which a technology or vendor can still be swapped, because early adoption compounds.
Inspect evidence · source links · passage review pending Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Claims and evidence · 3 synthesis Confidence: high Contestability: medium
Accountability depends on an actor's role, the context of use, and its practical ability to act; it is not a single obligation applied identically to everyone.
Our reading: The general principle comes from OECD; NIST illustrates organizational role allocation. This synthesis does not make the voluntary framework a legal duty.
Inspect evidence · 2 passages Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
OECD AI Principles (opens in a new tab) Principle 1.5 — accountability; first three paragraphs
Roles and context shape accountability. Traceability supports inquiry; risk management should be ongoing and proportionate to an actor’s ability to act.
Passage checked 2026-10-11 · oecd-accountability
interpretation Confidence: medium Contestability: medium
Many governance failures begin before any rule is broken: competitive pressure, funding, procurement, liability exposure and performance metrics make some behavior rational regardless of stated principles.
Our reading: The framework memo's position. The UN report supplies the race-to-the-bottom concern at the international level; NIST names incentive structures and culture as conditions for effective risk management (section 1.2.4).
Inspect evidence · 2 passages Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Governing AI for Humanity: Final Report (opens in a new tab) Executive summary, para. xvi (p. 8)
Governance regimes are said to need global span to avert AI arms races or a race to the bottom on safety and rights, to detect and respond to incidents spanning multiple jurisdictions, and to share AI’s benefits.
Scope note: An advisory body report representing a majority consensus of its members in their personal capacity. It is not a UN member-state decision. Page locators refer to the PDF's printed page numbers.
Passage checked 2026-09-06 · un-hlab-race-to-bottom
source-assertion Confidence: high Contestability: low
A developer withdrew an assistant update that had become overly flattering and agreeable, attributing the problem to training that weighted short-term user feedback too heavily.
Our reading: The developer’s own account; it shows the incentive is real inside development, but its explanation has not been independently verified.
Inspect evidence · 1 passage Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Sycophancy in GPT-4o: What happened and what we’re doing about it (opens in a new tab) Opening sections
The developer rolled back an update that made its assistant overly flattering or agreeable, attributing it to training that focused too much on short-term user feedback, and listed changes to training, testing and user controls.
Scope note: A developer’s account of its own product incident; not independently verified.
Passage checked 2026-10-18 · openai-rollback