Governments should fund open-source AI safety tools and red-teaming infrastructure as public goods.

public-interest-ai research-policy ai-governance ai-policy ai-regulation gov ai-safety open-source policy

Share this with your representatives, journalists and friends

Key insights

Across the quotes, evaluation capacity itself is repeatedly treated as part of AI safety, not just as a later compliance activity.
Open sourcing is usually defended as a way to increase auditability, reproducibility, shared learning, and access to rigorous testing, rather than as an end in itself.
The same call for public funding is justified both by security concerns, such as cyberattack simulation and national-risk assessment, and by equity concerns, such as local-language support and avoiding concentrated oversight.
Many sources describe concrete public institutions and tools already being funded or operated, including safety institutes, research labs, evaluation platforms, and national-lab testbeds.
Coordination is a recurring theme, with governments, standards bodies, universities, companies, and international partners described as co-builders or co-users of this infrastructure.

Arguments for

Government-led evaluation capacity

Supporters argue that the most advanced AI systems raise public safety and national security questions that require technical expertise, resources, and access that no single state or private actor can fully replicate. They therefore call for government-led testing, third-party assessment capacity, transparent evaluations, and publicly funded organizations able to track emerging risks.

First, the federal government should lead the testing and evaluation of the most advanced systems. Frontier AI raises national security and public safety questions that require technical expertise, resources, and access that no state can fully replicate.

OpenAI · Jul 15, 2026 · full quote
The Commission will help establish an EU evaluation capacity to strengthen third-party assessment of AI capabilities and risks globally, supporting the regulatory function of the AI Office.

European Commission · Jul 7, 2026 · full quote
Canada will invest $50 million to expand the capabilities of the Canadian AI Safety Institute to track emerging AI risks, advance technical research, and to conduct transparent evaluations of AI models.

Innovation, Science and Economic Development Canada · Jun 8, 2026 · full quote
The Government has built in response: an Australian AI Safety Institute, up and running, staffed with world-class expertise, and already inside the engine rooms of frontier AI, testing the most powerful models on the planet.

Andrew Charlton · Jul 7, 2026 · full quote
I also think we need some regulation and organizations that are directly funded by the government to provide evaluations of these systems.

Jacob Tsimerman · Jul 31, 2026 · full quote

Open and auditable tools

Another argument is that safety tools should not remain exclusive to the largest companies. Open tools, methods, and infrastructure are said to make evaluation more transparent, auditable, reproducible, verifiable, and easier for organizations of different sizes to adopt without starting from zero.

Safety tools cannot remain exclusive to the largest companies; they must be open, auditable, and collectively improved—serving as a common good for the benefit of society. This is how we will build trustworthy AI that truly protects our children, our teenagers, and all our fellow…

Anne Le Hénanff · Jun 2, 2026 · full quote
By open-sourcing not only Inspect, but also our methods and infrastructure where we can, we hope to help researchers, organisations, and other evaluators stand up rigorous evaluation capability without starting from zero. We hope that the Engineering Playbook provides researchers…

The AI Security Institute · Jun 18, 2026 · full quote
The UK, through AISI, has developed the world’s most widely used government-backed evaluation tools. Inspect, InspectSandbox, InspectCyber, and our latest release, ControlArena, are now being used by governments, companies, and academics around the world. These open tools lower t…

Kanishka Narayan · Feb 10, 2026 · full quote
By building safety tools in the open, this partnership will help organizations of every size adopt AI more responsibly. It strengthens trust, supports Canadian leadership in safe AI, and helps ensure the safeguards people rely on are transparent, verifiable, and shared.

Evan Solomon · Jun 2, 2026 · full quote
Project Moonshot is one of the world’s first Large Language Model (LLM) evaluation toolkits. It is an open-source solution that combines benchmarking and red-teaming, enabling developers and compliance teams to test and evaluate LLMs and their applications effectively.

Smart Nation Singapore · May 21, 2026 · full quote

Underprovided risky ecosystems

Several quotes describe open-source and open-weight AI ecosystems as both public goods and distinctive security challenges. Because private incentives may underprovide maintenance and safety work, and openness can bring risks such as adversarial attacks, malicious modification, and supply-chain compromise, governments are urged to fund scanners, robustness research, disclosure practices, and ecosystem monitoring.

The challenge is especially acute for open source software, since it’s a public good. Open source software underpins every software service we rely upon, including across critical infrastructure and the Federal government. A string of recent incidents have highlighted the need fo…

Jack Cable · Jun 4, 2026 · full quote
Corporate AI often prioritizes profit over public good, and critical capabilities (safety, security, equitable access) may never be developed privately.

Bruce Schneier · 2026 · full quote
Recognizing that their inherent openness introduces unique security risks, the US government should prioritize research into identifying and mitigating vulnerabilities in open-weight AI models. This includes funding research on adversarial robustness including attacks and testing…

IEEE-USA · Apr 20, 2026 · full quote
The Committee directs NIST to conduct a multi-stakeholder process to examine security risks associated with open-source artificial intelligence models such as model integrity, malicious modification, supply chain compromise, and data leakage. As part of this process, NIST shall i…

House Committee on Appropriations · May 2026 · full quote
Safety for open-weight models whose capabilities are nearing frontier closed models requires unique tools for open-weight model safety and ecosystem monitoring. [...] Accelerated public and private investment in research is required to keep pace with commercially driven growth in…

The 2026 Singapore Consensus on Global AI Safety Research Priorities · Jul 2026 · full quote
For organizations deploying open-weight models, a critical question emerges: how do you know the model you downloaded is safe to run? We believe defensive security tools should be widely available. AMS represents our contribution to a safer AI ecosystem—one where developers every…

Glen Messenger · Apr 27, 2026 · full quote

Shared standards and testbeds

Many supporters emphasize common datasets, benchmarks, attack simulators, cyber ranges, testbeds, and standards as shared infrastructure for safety work. They argue that this connective tissue can reduce duplicated effort, guide research, and let government, academia, and industry evaluate systems in more consistent ways.

Companies and governments should invest in shared open infrastructure for AI defense — datasets, evaluation frameworks, attack simulators and red-teaming tools — much as past generations invested in open source software.

NVIDIA · Jul 27, 2026 · full quote
GOLD EAGLE, established in President Trump’s June 2, 2026 Executive Order “Promoting Advanced Artificial Intelligence Innovation and Security” (EO 14409), represents a new operational model for cyber defense. This new model will leverage frontier AI capabilities to continue advan…

The White House · Jul 14, 2026 · full quote
At the European level, the existing legal frameworks, including NIS2, CRA and the EU AI Act should be leveraged, to ensure that systemic risks stemming from the most advanced AI models are assessed and mitigated. To this end, it may be useful to establish EU-wide state-of-the-art…

European Union Agency for Cybersecurity (ENISA) · Jul 2026 · full quote
This legislation brings together private sector and government experts to develop voluntary standards for AI, create new assessment tools, and conduct testing that will ensure the United States leads in AI-driven innovation and competitiveness for decades to come. The NIST Center…

Maria Cantwell · Feb 26, 2026 · full quote
Moving forward, the program will establish a forum comprised of universities, industry, and U.S. government representatives to fund, guide, and manage fast-paced, university-led research projects.

Defense Advanced Research Projects Agency (DARPA) · Jun 1, 2026 · full quote
GSA is preparing joint evaluation efforts with the National Institute of Standards and Technology’s AI Safety Institute and the Cybersecurity and Infrastructure Security Agency’s red-teaming initiative to benchmark frontier models for robustness, adversarial resistance, and bias.

U.S. General Services Administration · May 11, 2026 · full quote

Distributed capacity and equity

Some advocates stress that evaluation capacity should be publicly financed across regions, with local-language benchmarks, regionally hosted infrastructure, and shared platforms. They present this as a way to avoid outsourcing oversight to the same few actors, build strategic autonomy, and support cooperation across countries with different contexts.

We call for publicly financed, regionally hosted capacity to evaluate AI systems before they are deployed, and open benchmarks built with and for under-resourced languages, so oversight is not outsourced to the same few actors it is meant to oversee.

Rebecca Rektimbo · Jul 7, 2026 · full quote
The report advances ten recommendations under the SAFE framework — Standards, Accountability, Finance, and Equity — including adopting the Model Openness Framework as a shared reference for assessing openness, building governance support around the Digital Public Goods Alliance e…

United Nations Office for Digital and Emerging Technologies · Jun 25, 2026 · full quote
We must therefore treat the capacity to evaluate and govern AI as an integral part to safety itself, supported through shared tools, pooled expertise, and cooperation across regions.

Government of Pakistan · Jul 7, 2026 · full quote
In order to reinforce the warning’s message, the General Board calls on the EU to scale up its capacity, expertise and strategic autonomy in this critical area. This requires a coordinated answer of all parties, including AI providers, software providers, security firms, open-sou…

European Systemic Risk Board · Jul 7, 2026 · full quote
The summit provides a crucial platform to establish a shared digital platform that consolidates tools, frameworks, and benchmarks to advance responsible AI development and deployment across South Asia. An AI Safety Commons would serve as both a technical repository and an operati…

Amlan Mohanty · Feb 2026 · full quote

Middle ground

Controlled-access shared infrastructure

Some supportive quotes envision public-good infrastructure with controlled or targeted access rather than unrestricted release. They describe shared red-teaming or testbed capacity that is subsidized, cost-recoverable, or provided in secure environments, especially for SMEs, state and local governments, and pre-deployment testing of proprietary models.

If the modifier is required to conduct Step 3 as part of their initial burden when Steps 1 and 2 are inconclusive, the oversight body may, on application by an SME modifier, provide access to a shared government-funded red-teaming infrastructure at cost, recoverable from the ulti…

Bipin Kumar B · Mar 23, 2026 · full quote
Making available a research test-bed to private sector and academic researchers, on a subsidized basis, to engage in artificial intelligence security research, including through the secure provision of access in a secure environment for pre-deployment testing of proprietary third…

United States Senate Select Committee on Intelligence · May 20, 2026 · full quote
Particularly in the context of AI-driven vulnerability discovery and exploit development, state and local governments will need ready access to the tools and capabilities that allow them to discover and remediate technical vulnerabilities in their systems. CISA and the MS-ISAC an…

Samir Jain · May 21, 2026 · full quote

Within the provided quotes, the main variation is not over whether governments should play a role, but over how that role should be organized: through open tools, national institutes, common benchmarks, secure testbeds, or regionally distributed capacity. The cited sources converge on a public role in AI evaluation and red-teaming while raising practical design questions about openness, access controls, and cross-sector coordination.

Cast your vote:

Results (86 votes):

Total (86 votes)

For 86 (100%) Abstain 0 (0%) Against 0 (0%)

For (86)

Urvashi Aneja
Founder of research organization Digital Futures Lab

I am this person! Add a sourced quote Report comment

votes For and says:

But if governments don’t invest in the safety infrastructure now, public trust will erode, and the opportunity to leverage benefits from AI will go away, and then we’re looking at deepening inequality.
AI Verified source (Sep 1, 2026)

added 19d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Jacob Tsimerman
Mathematician, Fields Medalist, and professor at the University of Toronto.

I am this person! Add a sourced quote Report comment

votes For and says:

I also think we need some regulation and organizations that are directly funded by the government to provide evaluations of these systems.
AI Verified source (Jul 31, 2026)

added 15d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Ministry of Electronics and Information Technology
Government AI policy institution.

I am this person! Add a sourced quote Report comment

votes For and says:

The Safe & Trusted AI pillar of the IndiaAI Mission aims to promote the responsible development, deployment and adoption of AI through indigenous governance frameworks, standards, tools and evaluation mechanisms. 13 Responsible AI projects have been ...
more AI Verified source (Jul 27, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Government of India
National government of India, speaking through its Permanent Representative at the United Nations on artificial intelligence governance.

I am this person! Add a sourced quote Report comment

votes For and says:

The Safe & Trusted AI pillar of the IndiaAI Mission aims to promote the responsible development, deployment and adoption of AI through indigenous governance frameworks, standards, tools and evaluation mechanisms. 13 Responsible AI projects have been ...
more AI Verified source (Jul 27, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
NVIDIA
GPU and AI computing company

I am this person! Add a sourced quote Report comment

votes For and says:

Companies and governments should invest in shared open infrastructure for AI defense — datasets, evaluation frameworks, attack simulators and red-teaming tools — much as past generations invested in open source software.
AI Verified source (Jul 27, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Advanced Research and Invention Agency (ARIA)
United Kingdom public R&D agency sponsoring AI-safety research and open-source coordination infrastructure.

I am this person! Add a sourced quote Report comment

votes For and says:

To kickstart Phase 1 of this programme, we will be funding teams to develop open-source coordination infrastructure and perform fundamental research that moves us from empirical to theory-driven guarantees in agentic coordination.
AI Verified source (Jul 18, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
OpenAI
AI research organization

I am this person! Add a sourced quote Report comment

votes For and says:

First, the federal government should lead the testing and evaluation of the most advanced systems. Frontier AI raises national security and public safety questions that require technical expertise, resources, and access that no state can fully replic...
more AI Verified source (Jul 15, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Chris Lehane
OpenAI chief global affairs officer

I am this person! Add a sourced quote Report comment

votes For and says:

First, the federal government should lead the testing and evaluation of the most advanced systems. Frontier AI raises national security and public safety questions that require technical expertise, resources, and access that no state can fully replic...
more AI Verified source (Jul 15, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
The White House
Executive office of U.S. president

I am this person! Add a sourced quote Report comment

votes For and says:

GOLD EAGLE, established in President Trump’s June 2, 2026 Executive Order “Promoting Advanced Artificial Intelligence Innovation and Security” (EO 14409), represents a new operational model for cyber defense. This new model will leverage frontier AI ...
more AI Verified source (Jul 14, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Rebecca Rektimbo
Connectivity Technical Projects Coordinator for LOCNET, a collaboration initiative of the Association for Progressive Communications and Rhizomatica, speaking at the UN Global Dialogue on AI Governance.

I am this person! Add a sourced quote Report comment

votes For and says:

We call for publicly financed, regionally hosted capacity to evaluate AI systems before they are deployed, and open benchmarks built with and for under-resourced languages, so oversight is not outsourced to the same few actors it is meant to oversee.
AI Verified source (Jul 7, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
European Systemic Risk Board
European Union body responsible for macro-prudential oversight and the prevention or mitigation of systemic risks to financial stability.

I am this person! Add a sourced quote Report comment

votes For and says:

In order to reinforce the warning’s message, the General Board calls on the EU to scale up its capacity, expertise and strategic autonomy in this critical area. This requires a coordinated answer of all parties, including AI providers, software provi...
more AI Verified source (Jul 7, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Andrew Charlton
Australian Assistant Minister for Science, Technology and the Digital Economy

I am this person! Add a sourced quote Report comment

votes For and says:

The Government has built in response: an Australian AI Safety Institute, up and running, staffed with world-class expertise, and already inside the engine rooms of frontier AI, testing the most powerful models on the planet.
AI Verified source (Jul 7, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
European Commission
EU's executive body

I am this person! Add a sourced quote Report comment

votes For and says:

The Commission will help establish an EU evaluation capacity to strengthen third-party assessment of AI capabilities and risks globally, supporting the regulatory function of the AI Office.
AI Verified source (Jul 7, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Government of Pakistan
National government of Pakistan, speaking through its delegation at the UN Global Dialogue on AI Governance in July 2026.

I am this person! Add a sourced quote Report comment

votes For and says:

We must therefore treat the capacity to evaluate and govern AI as an integral part to safety itself, supported through shared tools, pooled expertise, and cooperation across regions.
AI Verified source (Jul 7, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Xinchen Hua Lu
UN AI panel member and researcher at CSIRO Australia.

I am this person! Add a sourced quote Report comment

votes For and says:

The government industrial research organizations, they could share their current evaluation tools or methods, evidence through the common platforms.
AI Verified source (Jul 7, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
European Union Agency for Cybersecurity (ENISA)
European Union agency dedicated to achieving a high common level of cybersecurity across Europe.

I am this person! Add a sourced quote Report comment

votes For and says:

At the European level, the existing legal frameworks, including NIS2, CRA and the EU AI Act should be leveraged, to ensure that systemic risks stemming from the most advanced AI models are assessed and mitigated. To this end, it may be useful to esta...
more AI Verified source (Jul 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
The 2026 Singapore Consensus on Global AI Safety Research Priorities
July 2026 international consensus report on AI safety research priorities produced by over 100 contributors from 13 countries.

I am this person! Add a sourced quote Report comment

votes For and says:

Safety for open-weight models whose capabilities are nearing frontier closed models requires unique tools for open-weight model safety and ecosystem monitoring. [...] Accelerated public and private investment in research is required to keep pace with...
more AI Verified source (Jul 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Baroness Lloyd of Effra
Liz Lloyd; UK Parliamentary Under-Secretary of State for AI and Digital Government (DSIT)

I am this person! Add a sourced quote Report comment

votes For and says:

Dedicated AIRR capacity supports the AI Security Institute and the Sovereign AI Unit, including guaranteed resource for mission-critical AI safety research, and strategically selected UK AI startups.
AI Verified source (Jun 29, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Jan Wehner
AI governance researcher at the Institute for AI Policy and Strategy, writing on AI safety and cybersecurity.

I am this person! Add a sourced quote Report comment

votes For and says:

To avoid losing visibility into cutting-edge AI cyberattack capabilities exactly when it matters most, I argue governments should fund independent threat-intelligence collection and sharing, develop AI-specific forensic capabilities, build the capaci...
more AI Verified source (Jun 29, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
United Nations Office for Digital and Emerging Technologies
United Nations office supporting global digital cooperation and AI governance.

I am this person! Add a sourced quote Report comment

votes For and says:

The report advances ten recommendations under the SAFE framework — Standards, Accountability, Finance, and Equity — including adopting the Model Openness Framework as a shared reference for assessing openness, building governance support around the D...
more AI Verified source (Jun 25, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
University of Oxford
University and research institution publishing official information on government-backed AI research and safety.

I am this person! Add a sourced quote Report comment

votes For and says:

The University of Oxford has been selected to host a new, cross-university government-backed UK AI Research Lab. The new lab is part of a £60 million investment announced today by the UKRI's Engineering and Physical Sciences Research Council (EPSRC) ...
more AI Verified source (Jun 23, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Matthew Ferren
Council on Foreign Relations International Affairs Fellow in National Security and former U.S. Department of Defense cyber-policy official.

I am this person! Add a sourced quote Report comment

votes For and says:

When the cost of providing access to AI models conflicts with corporate profit concerns, the U.S. government should step in by providing targeted funding and facilitating cross-lab coordination and efficiency. The U.S. government should also facilita...
more AI Verified source (Jun 22, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
AI Security Institute (AISI)
UK government AI safety institute

I am this person! Add a sourced quote Report comment

votes For and says:

By open-sourcing not only Inspect, but also our methods and infrastructure where we can, we hope to help researchers, organisations, and other evaluators stand up rigorous evaluation capability without starting from zero.
AI Verified source (Jun 18, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
The AI Security Institute
UK government research organisation focused on AI safety and security

I am this person! Add a sourced quote Report comment

votes For and says:

By open-sourcing not only Inspect, but also our methods and infrastructure where we can, we hope to help researchers, organisations, and other evaluators stand up rigorous evaluation capability without starting from zero. We hope that the Engineering...
more AI Verified source (Jun 18, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Mischa Spiegelmock
Technical staff member at METR working on frontier model evaluations

I am this person! Add a sourced quote Report comment

votes For and says:

Absolutely central to METR's evaluations is Inspect, which we've built our own open source cloud eval platform on top of, along with actively contributing to the Inspect codebase to run larger and more complex evaluations.
AI Verified source (Jun 18, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
JJ Allaire
Founder and Executive Chairman of Posit; creator of Inspect

I am this person! Add a sourced quote Report comment

votes For and says:

I created Inspect because I wanted to provide a strong technical foundation for governments, researchers, and labs to collaborate on testing and understanding frontier models. I am thrilled to have seen so many organisations adopt it, as well as to h...
more AI Verified source (Jun 18, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Innovation, Science and Economic Development Canada
Government of Canada department responsible for innovation, science, and economic development.

I am this person! Add a sourced quote Report comment

votes For and says:

Canada will invest $50 million to expand the capabilities of the Canadian AI Safety Institute to track emerging AI risks, advance technical research, and to conduct transparent evaluations of AI models.
AI Verified source (Jun 8, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Jack Cable
CEO and co-founder of Corridor; former CISA Secure by Design initiative lead and cybersecurity expert.

I am this person! Add a sourced quote Report comment

votes For and says:

The challenge is especially acute for open source software, since it’s a public good. Open source software underpins every software service we rely upon, including across critical infrastructure and the Federal government. A string of recent incident...
more AI Verified source (Jun 4, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Donald Trump
President of the United States

I am this person! Add a sourced quote Report comment

votes For and says:

Within 30 days of the date of this order, the Director of OMB, in coordination with the National Cyber Director and the Director of CISA, shall determine whether any Federal grant programs have available and relevant funding that can be directed towa...
more AI Verified source (Jun 2, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Camille François
President and CEO of Robust Open Online Safety Tools (ROOST)

I am this person! Add a sourced quote Report comment

votes For and says:

The safety problems that matter most are often the ones no one has been able to solve in the open. Safety guardrails for youth resilience is a clear example: the need is global, but the tools are scarce.
AI Verified source (Jun 2, 2026)

added 2mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate