EDITORIAL: Washington Negotiates Voluntary AI Standards—A Necessary Gamble

Clearer Testing Criteria

According to the Financial Times, the new standards under negotiation would establish technical benchmarks for advanced models as well as specific timelines, while clarifying who can access these systems in the United States and abroad. This clarification would directly address the criticism voiced in recent weeks regarding the opacity of the process.

Google is also reportedly in discussions with the government ahead of the release of its own advanced coding models, whose cybersecurity capabilities are said to exceed those of previous generations, according to the British daily.

A Delicate Balance Between Security and Competitiveness

The challenge for Washington is to establish sufficient safeguards to prevent the cybersecurity and biosecurity risks associated with the most powerful models, without stifling American innovation in the face of increasingly fierce international competition, particularly from China.

I believe the West does not have the luxury of choosing between security and speed: it must deliver both simultaneously, because every month of poorly calibrated regulatory delay directly benefits Chinese labs, which have no qualms about accelerating their efforts without burdening their teams with such precautions.

OpenAI and the GPT-5.6 Case

OpenAI announced last week that it would present its new GPT-5.6 model—which includes the Sol, Terra, and Luna variants—only to a “small group of trusted partners” prior to any public rollout, at the request of the U.S. government and for cybersecurity reasons, according to TechCrunch. CEO Sam Altman reportedly told his employees that the government would “approve access on a client-by-client basis” during the preview period.

This situation provides a concrete example of how the executive order is already being implemented, well before the new framework of voluntary standards currently under negotiation has even been finalized.

Anthropic and the Precedent of Export Controls

The Anthropic case highlights the limitations of the current system: its Claude Mythos and Fable 5 models were subject to export controls for several weeks before the Department of Commerce lifted these restrictions in late June, following a joint effort to assess the risks, according to CNN and Commerce Secretary Howard Lutnick.

I see the Anthropic episode as proof that the system works, even if imperfectly: a problem was identified, addressed, and then resolved through cooperation rather than confrontation—which is exactly the kind of process that a more formalized framework should make possible to replicate more quickly in the future.

A voluntary initiative already in place in 2023

The idea of voluntary standards for AI is not entirely new: in July 2023, the Biden administration had already secured non-binding commitments from seven major companies—including Amazon, Google, Meta, Microsoft, and OpenAI—specifically regarding external safety testing prior to release, according to Politico and the American National Standards Institute.

This precedent demonstrates that the principle of voluntary participation has been a consistent feature across successive U.S. administrations, regardless of their political affiliation—proof that the issue extends far beyond the usual partisan divides.

What Sets the Current Approach Apart

Unlike the largely symbolic 2023 framework, the 2026 executive order is accompanied by concrete enforcement mechanisms, notably through NIST’s Center for AI Standards and Innovation, which is already conducting pre-deployment assessments with leading U.S. laboratories, according to The GPU Trade.

I believe this difference in rigor—between a superficial commitment in 2023 and a genuine technical evaluation mechanism in 2026—marks a welcome maturation in the way Washington approaches the governance of artificial intelligence.

The AI Race as a Battleground for Strategic Rivalry

The June executive order had initially been delayed due to concerns about its potential impact on U.S. competitiveness vis-à-vis China, according to The Hill—a reminder that AI regulation cannot be considered in isolation from the global technological rivalry between major powers.

The executive order also explicitly calls for strengthening national cyberdefense capabilities and facilitating access to defensive AI tools for critical infrastructure, from rural hospitals to community banks.

Why the West Must Stay Ahead

This technological race is not merely a commercial issue: it directly shapes the balance of power between the West and its strategic rivals, foremost among them China, which is investing heavily in its own artificial intelligence capabilities without the same transparency requirements.

I remain convinced that the U.S. advantage in artificial intelligence is one of the most valuable pillars of Western power today, and that a well-calibrated regulatory framework—neither too lax nor too rigid—is precisely what will ensure its long-term preservation.

Uncertainty Surrounding Decision-Making Criteria

Several analysts, including commentator Ball, as cited by The Hill, point out that “literally no one knows” what security standards a company must meet to obtain authorization for the widespread deployment of its models—a lack of clarity that, in practice, amounts to a systematic rejection until the criteria are formalized.

This uncertainty directly impacts the planning of artificial intelligence labs, which must navigate changing requirements without clear visibility into timelines or final approval criteria.

Transparency Still Insufficient, According to Some Experts

Critics, including those within the companies themselves, are calling for the publication of testing criteria, a list of trusted partners, and a mechanism for appeal in case of disagreement—all elements that remain absent from the current framework, according to several specialized legal analyses.

I believe these criticisms are well-founded and deserve to be heard: a voluntary framework can only function sustainably if it is based on public and predictable rules; otherwise, it risks becoming nothing more than an informal power struggle between Washington and the major tech companies.

A Joint Call for Predictability

OpenAI and Anthropic have both called for a federal framework establishing clear safeguards around the highest risks associated with artificial intelligence, while disagreeing on the extent of the role the government should play in the development process itself, according to The Hill.

This joint call for predictability reflects a simple industry reality: labs prefer binding rules that are known in advance rather than an unpredictable approval process that delays their launch schedules without public justification.

The Promise of a Repeatable Process

According to TechCrunch, OpenAI has already indicated its willingness to work with the administration to establish a repeatable process for future model releases—a way to transform a one-off, ad hoc experience into a standardized mechanism applicable across the entire industry.

I believe this stated commitment to building a repeatable process—rather than being subject to case-by-case decisions—is the most encouraging sign in this entire sequence: it is exactly the kind of structured cooperation the West needs to remain competitive without sacrificing security.

A Welcome Clarification, Despite Some Uncertainties

The voluntary framework currently being negotiated between the White House, OpenAI, Google, and Anthropic represents a serious attempt to structure a technological field whose rapid evolution far exceeds the ability of traditional institutions to regulate it effectively and in a timely manner.

It remains to be seen whether the announcement expected in the coming days will finally provide the clarity sought by the companies involved, or whether it will merely formalize a status quo that has already been widely criticized for its lack of transparency.

What the West Has to Lose or Gain

Against a backdrop in which China continues to invest heavily in its own artificial intelligence capabilities, the United States’ ability to establish a framework that is both secure and flexible enough to preserve innovation will be decisive for the global technological balance in the coming years.

I conclude this editorial convinced that an imperfect but transparent framework is always better than constant improvisation, and that the West’s future credibility in the race for artificial intelligence will depend largely on its ability to provide this clarification quickly and effectively.

Signed, Maxime Marquette, columnist