AI companies are constructing their own internal arms control infrastructure to prevent language models from helping synthesize chemical weapons or build explosives, but these voluntary measures lack any external accountability or enforcement mechanism. Anthropic and OpenAI have posted job listings for policy managers focused on chemical and biological weapons risks, roles that would traditionally exist at defense ministries or international security bodies. The problem is clear: without mandatory external oversight, these internal safety commitments will collapse under competitive pressure, according to policy experts analyzing the companies' own statements .

Why Are AI Companies Hiring Weapons Experts?

When Anthropic posted a job listing for a "Policy Manager, Chemical Weapons and High-Yield Explosives," it signaled something policymakers have been slow to acknowledge directly. The person hired for this position will design evaluation methodologies to assess what Claude, Anthropic's AI assistant, can do in domains like chemical weapons synthesis and explosives development. They will develop strategies to identify and mitigate potential misuse in model outputs and run rapid-response protocols when the company detects escalating queries in these categories .

OpenAI reportedly posted a near-identical role for a researcher focused on biological and chemical risks. These are not typical compliance hires. In functional terms, they are performing the work of verification officers at international bodies like the Organisation for the Prohibition of Chemical Weapons (OPCW), which has 193 member states and a verification regime backed by international law. The difference is critical: traditional verification officers operate within treaty frameworks and answer to governments and international institutions. These new hires answer to a board of directors and a CEO navigating a competitive commercial market .

The urgency is real. In March 2026, the OPCW released a report on how AI intersects with the Chemical Weapons Convention, noting that AI-enabled tools are already transforming chemical research. Molecular modeling and AI-assisted synthesis planning now allow researchers to identify chemical pathways faster and with less expertise than was previously required. Predictive toxicity analysis has accelerated the process further, lowering the barrier to designing harmful chemicals that could be weaponized .

What's the Gap Between AI Companies and International Bodies?

The OPCW wants AI companies to engage with it. The organization is building capacity-building programs, consulting with industry and academia, and pushing member states to develop shared norms. What it cannot do is compel a private company in San Francisco to submit its model evaluations for external review or disclose the results of internal red-teaming. It has no mechanism to require coordination with its technical secretariat before deploying a new model version. That gap, between an international body with deep domain expertise and no authority over AI companies, and AI companies with enormous capabilities and no international obligations, is where the world currently stands .

The problem became even more apparent when Anthropic released version 3.0 of its Responsible Scaling Policy in February 2026. The document is detailed and serious, including new requirements for periodic Risk Reports, external expert review in high-risk scenarios, whistleblower protections, and a Frontier Safety Roadmap that the company commits to publicly grading itself. But the update drew attention primarily for what it removed. Previous versions included a binding commitment: if Anthropic could not demonstrate adequate safety measures before crossing a capability threshold, it would pause development. That commitment is gone, replaced by voluntary public goals the company describes as ambitious but explicitly non-binding .

"We didn't really feel, with the rapid advance of AI, that it made sense for us to make unilateral commitments if competitors are blazing ahead," stated Jared Kaplan, Chief Science Officer at Anthropic.

Jared Kaplan, Chief Science Officer at Anthropic

Anthropic's reasoning was candid. Unilateral safety commitments do not work if competitors are not making equivalent ones. A company that stops training while others continue does not make the world safer; it cedes market position and, eventually, the ability to shape the technology at all. The only durable solution, the company acknowledged, is coordination through external rules that apply to everyone .

How to Strengthen AI Weapons Safeguards: What Experts Recommend

• Mandatory External Disclosure: Require frontier AI developers to disclose model evaluations related to chemical, biological, and radiological weapons to external bodies with real authority, not just publish voluntary risk frameworks.
• International Treaty Framework: Establish binding international agreements similar to existing arms control regimes, with verification mechanisms that apply to all AI companies regardless of jurisdiction or competitive position.
• Regulatory Formalization: Expand requirements like California's SB-53 and the EU AI Act to specifically address weapons-domain evaluations, moving beyond voluntary disclosure to mandatory third-party review before model deployment.

The Centre for the Governance of AI at Oxford drew the logical conclusion in its analysis of Anthropic's updated policy: "If the core problem is collective action, Anthropic should push for stronger regulation, according to its own logic." The report added that while Anthropic appears to be taking some steps in this direction, "its efforts seem to lag behind what its own logic suggests" .

Even within the voluntary safety infrastructure that exists, coverage is uneven. Frontier AI safety work has concentrated heavily on pandemic-scale biological risks, modeling scenarios in which a lone actor uses AI to engineer a pathogen capable of mass casualties. That is a legitimate threat, but it has drawn attention away from others. Chemical weapons, improvised explosive attacks, and radiological devices are lower-profile than a pandemic but considerably more accessible to a motivated actor and receive substantially less systematic attention .

Labs publish evaluations of whether their models could enable a pandemic; they do not typically detail whether those same models could assist in a chemical attack or help someone circumvent export controls on precursor materials. The hiring of dedicated chemical weapons policy managers at two major labs suggests the companies have registered this concern. What is less clear is whether the evaluations those managers conduct will ever be visible to anyone outside the companies .

The 2026 International AI Safety Report, a multi-institutional effort coordinated through the UK AI Safety Institute, found that most risk-management practices at frontier labs remain voluntary. A handful of jurisdictions have begun formalizing limited requirements. California's SB-53 and elements of the EU AI Act now require frontier developers to publish risk frameworks; New York's RAISE Act will add similar obligations when it takes effect in 2027. None specifically addresses the weapons-domain evaluations that Anthropic and OpenAI are now staffing internally. What existing law requires, publishing a risk framework, and what the risk actually demands, mandatory disclosure of weapons-domain evaluations to external bodies with real authority, are not close to each other .

The companies best positioned to design workable AI governance frameworks are the ones that understand the technology most deeply. They are also the companies with the strongest competitive incentive to avoid any framework that constrains them more than their rivals. They can diagnose the problem clearly. Acting on the diagnosis is harder.