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 · whole model
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…
actor AI developer or provider actor Deploying organization actor Affected people actor Civil society and independent researchers Incentives …the incentives…
incentive Deployment pressure incentive Competitive race dynamics incentive Liability exposure incentive Procurement leverage
Institutions …institutions…
institution Regulator or public authority institution Internal safety and governance team institution Standards body institution International institution
Infrastructures …infrastructures…
infrastructure Evaluation capacity infrastructure Compute infrastructure Data, language resources, and data labor infrastructure Energy, water, and minerals infrastructure Data work and content moderation Safeguards …and safeguards around AI…
safeguard Risk management process safeguard Impact assessment safeguard Traceability safeguard Audit record safeguard Meaningful human oversight safeguard Supervision and regulatory review safeguard Pre-deployment evaluation and red teaming safeguard Incident reporting and post-deployment monitoring safeguard Complaint and appeal routes safeguard Review and sunset provisions safeguard Compute controls Benefits and harmful patterns …so that its benefits are broadly shared and harmful patterns are addressed…
principle Accountability principle Safety principle Broadly shared benefit pattern Contestable decision pattern Concentration of power pattern Systematically unreliable provenance pattern Over-reliance and deskilling pattern Sycophancy and engagement-driven agreement 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
Safeguard · Safeguards contested
Compute controls Reporting thresholds, licensing, export controls, and access conditions applied to advanced compute, using its detectability and concentration as a point of intervention.
What does the control make visible or prevent, and whom does it consolidate power among?
Tests Power Distribution Adaptability
Lifecycle Model development Deployment
Levels National International Recorded relationships Read each statement from its first entity to its second. Status describes our mapping, not certainty of cause.
Compute — is the lever for → Compute controls
Mapping: asserted
Because compute is detectable, excludable and quantifiable, it can be the object of reporting thresholds, licensing and export controls in a way data and algorithms cannot.
Inspect evidence · 1 passage Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Computing Power and the Governance of Artificial Intelligence (opens in a new tab) Abstract
Relative to data and algorithms, AI-relevant compute is presented as a particularly effective point of intervention because it is detectable, excludable and quantifiable and is produced via an extremely concentrated supply chain. Policymakers could use it for regulatory visibility, resource allocation and enforcement; the authors warn that naive or poorly scoped compute governance carries significant risks to privacy, economic impacts and centralization of power, and suggest guardrails.
Scope note: A preprint by authors affiliated with AI developers, universities, and governance organizations; not peer reviewed. Several authors have institutional stakes in compute governance.
Passage checked 2026-09-06 · sastry-compute-abstract
Compute controls — may deepen → Concentration of power
Mapping: contested
Controls that only the largest actors can satisfy reduce access to dangerous capability while consolidating power among those already able to comply.
Inspect evidence · 1 passage Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Computing Power and the Governance of Artificial Intelligence (opens in a new tab) Abstract
Relative to data and algorithms, AI-relevant compute is presented as a particularly effective point of intervention because it is detectable, excludable and quantifiable and is produced via an extremely concentrated supply chain. Policymakers could use it for regulatory visibility, resource allocation and enforcement; the authors warn that naive or poorly scoped compute governance carries significant risks to privacy, economic impacts and centralization of power, and suggest guardrails.
Scope note: A preprint by authors affiliated with AI developers, universities, and governance organizations; not peer reviewed. Several authors have institutional stakes in compute governance.
Passage checked 2026-09-06 · sastry-compute-abstract
Compute controls — intended to protect → Safety
Mapping: asserted
The stated purpose of compute governance is regulatory visibility and enforcement against irresponsible or malicious development.
Inspect evidence · 1 passage Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Computing Power and the Governance of Artificial Intelligence (opens in a new tab) Abstract
Relative to data and algorithms, AI-relevant compute is presented as a particularly effective point of intervention because it is detectable, excludable and quantifiable and is produced via an extremely concentrated supply chain. Policymakers could use it for regulatory visibility, resource allocation and enforcement; the authors warn that naive or poorly scoped compute governance carries significant risks to privacy, economic impacts and centralization of power, and suggest guardrails.
Scope note: A preprint by authors affiliated with AI developers, universities, and governance organizations; not peer reviewed. Several authors have institutional stakes in compute governance.
Passage checked 2026-09-06 · sastry-compute-abstract
Claims and evidence · 2 source-assertion Confidence: high Contestability: medium
Compute is an unusually effective governance intervention point because it is detectable, excludable and quantifiable and comes from a concentrated supply chain; compute governance can also threaten privacy, economic outcomes, and centralize power.
Our reading: A preprint position, not a settled finding. Its authors include people at organizations that would operate or be subject to compute governance.
Inspect evidence · 1 passage Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Computing Power and the Governance of Artificial Intelligence (opens in a new tab) Abstract
Relative to data and algorithms, AI-relevant compute is presented as a particularly effective point of intervention because it is detectable, excludable and quantifiable and is produced via an extremely concentrated supply chain. Policymakers could use it for regulatory visibility, resource allocation and enforcement; the authors warn that naive or poorly scoped compute governance carries significant risks to privacy, economic impacts and centralization of power, and suggest guardrails.
Scope note: A preprint by authors affiliated with AI developers, universities, and governance organizations; not peer reviewed. Several authors have institutional stakes in compute governance.
Passage checked 2026-09-06 · sastry-compute-abstract
synthesis Confidence: medium Contestability: high
Concentration of power over AI's inputs and decisions is itself a governance problem, not only a side effect of unsafe systems.
Our reading: The UN advisory body treats concentration as a harm; the compute paper treats centralization as a risk of its own proposal. AIGovSpec's framework memo adopts the position that concentration is among the most important harmful patterns; that is an editorial stance.
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, paras. vii and xv (pp. 7–8)
The accelerating development of AI is said to concentrate power and wealth on a global scale, with geopolitical and geoeconomic implications. Equity is said to demand that more voices play meaningful roles in governance decisions, and the concentration of decision-making in the AI technology sector is described as unjustifiable.
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-concentration
Computing Power and the Governance of Artificial Intelligence (opens in a new tab) Abstract
Relative to data and algorithms, AI-relevant compute is presented as a particularly effective point of intervention because it is detectable, excludable and quantifiable and is produced via an extremely concentrated supply chain. Policymakers could use it for regulatory visibility, resource allocation and enforcement; the authors warn that naive or poorly scoped compute governance carries significant risks to privacy, economic impacts and centralization of power, and suggest guardrails.
Scope note: A preprint by authors affiliated with AI developers, universities, and governance organizations; not peer reviewed. Several authors have institutional stakes in compute governance.
Passage checked 2026-09-06 · sastry-compute-abstract