Global Policy

Technical AI Governance

We support governments and international organizations in developing technical AI governance regimes, regulating frontier AI systems, and establishing best practices.

highlights

Featured publications

November 24, 2025

The Loss of Control Playbook: Degrees, Dynamics, and Preparedness

Despite increasing policy and research attention to Loss of Control (LoC), decision- and policymakers are still operating in the absence of a uniform conceptualization and definition of LoC. Today, we bridge this gap through a novel taxonomy and preparedness framework for LoC that explores the degrees and dynamics of LoC through a comprehensive best-in-class literature review and presents actionable tools to counter relevant threats to national security and humanity.

Read more

Papers & Reports

April 17, 2025

AI Behind Closed Doors: a Primer on The Governance of Internal Deployment

In the race toward increasingly capable artificial intelligence (AI) systems, much attention has been focused on how these systems interact with the public. However, a critical blind spot exists in our collective thinking: the governance of highly advanced AI systems deployed within the frontier AI companies developing them. Today, we publish the first report of its kind providing a multi-faceted analysis of the risks and governance of internal deployment.

Read more

Papers & Reports

April 15, 2025

Capturing and Countering Threats to National Security: a Blueprint for an Agile AI Incident Regime

We propose a three-pronged approach to an AI incident regime that supports the establishment and implementation of a state capacity to track the national security threats that future frontier AI systems may pose and to swiftly execute countermeasures to contain and neutralize these threats.

Read more

Papers & Reports

February 6, 2025

Precursory Capabilities: A Refinement to Pre-deployment Information Sharing and Tripwire Capabilities

We explore methods to better operationalize existing AI capability thresholds through, what we call, ‘precursory capabilities'.

Read more

Papers & Reports

our findings

All research




Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

An Overview of Our Current Governance Efforts

Blog

policy recommendations, policy, workstreams, loss of control, LoC, internal deployment, dangerous capability evaluations, stakeholders, policy levers, governance team, technical governance, governments, AI policy, overview, portfolio

A Loss of Control Threat Map for AI Research and Development

Blog

threat map, AI R&D, R&D, automation, automating AI development, loss of control, LoC, pathways, threat model, automated AI R&D, risk map, self-improvement, interactive map

Misaligned AI as a New Insider Risk

Memorandums

insider risk, insider threat, policy memorandum, memo, high-stakes, national security, government agencies, contractors, classified, NISPOM, NDAA, coverage gap, sabotage, whistleblowing, blackmailing, misaligned, deployers, personnel security, security frameworks

The Need for Deeper, White-Box Access to Maintain State of the Art Evaluations for Loss of Control Threats

Blog

white-box access, white box, evaluation awareness, loss of control, LoC, evaluators, security assessments, model access, deeper access, interpretability access, evaluator access, internals

NIST RFI: Security Considerations for AI Agents

Submission

NIST, RFI, request for information, AI agents, agents, security, security considerations, comment, submission, public comment, US policy, standards, agent security, federal

Seven Provisions in the National Defense Authorization Act with High Potential to Accelerate AI Security

Blog

NDAA, National Defense Authorization Act, AI security, DoW, ODNI, CAISI, NIST, Intelligence Community, IC, AI Action Plan, national security, Senate, House of Representatives, assurance, provisions, defense bill, Congress, US legislation, defense

Internal Deployment of AI Models and Systems in the EU AI Act

Memorandums

EU AI Act, AI Act, EU, Europe, European Union, internal deployment, internal use, memorandum, memo, frontier AI companies, regulation, EU regulation, compliance, legal analysis

The Loss of Control Playbook: Degrees, Dynamics, and Preparedness

Papers & Reports

LoC, loss of control, taxonomy, preparedness framework, preparedness, national security, severity, persistence, threat map, DAP framework, policymakers, degrees, dynamics, playbook, humanity, literature review, report, escalation

Assurance of Frontier AI Built for National Security

Memorandums

assurance, national security, testing and evaluation, evaluation, memorandum, memo, acquisition, IC, Intelligence Community, DoW, frontier AI, defense, procurement, military AI, vetting

AI Behind Closed Doors: a Primer on The Governance of Internal Deployment

Papers & Reports

internal deployment, internally, frontier AI companies, governance, report, blind spot, risks, primer, behind closed doors, advanced AI systems, self-use, internal use, governance gap, frontier labs, lab governance