# I Tried to Understand America's New Framework for Advanced AI > According to the Financial Times, cited by several agencies including Reuters, the American government is reportedly in advanced talks with major artificial - Author: Maxime Marquette - Publisher: MadMax (https://mad-max.co) - Section: Profile - Published: 2026-07-05 - Canonical URL: https://mad-max.co/en/article/j-ai-essaye-de-comprendre-le-nouveau-cadre-americain-sur-l-ia-de-pointe ## Introduction: a file beyond my technical skill set ### An announcement that could change the technological landscape According to the **Financial Times**, cited by several agencies including **Reuters**, the American government is reportedly in advanced talks with major **artificial intelligence** companies to establish voluntary standards governing the release of new models, with a potential announcement as early as the week following **July 1, 2026** (**Reuters**). These standards would set benchmarks for advanced models and timelines, while clarifying who can access them inside the **United States** and abroad. My name is **Maxime Marquette**, and I have to be honest from the start: I am not an **artificial intelligence** engineer, and the technical subtleties of this file sometimes go over my head. But I believe a columnist can, and should, try to make these issues understandable for readers, even without mastering every technical detail. *I approach this subject with deliberate humility. I do not claim to understand every technical parameter of an advanced AI model, but I can recognize the geopolitical importance of this moment, and that is the angle I am focusing on.* ## The June 2 executive order, where this whole file begins ### A voluntary thirty-day window On **June 2, 2026**, President **Donald Trump** signed an executive order titled **Promoting Advanced Artificial Intelligence Innovation and Security**, creating a voluntary mechanism allowing the federal government to review the most powerful AI models before their public release (**The Guardian**). The order asks certain companies to voluntarily submit their new models for testing or evaluation roughly **thirty days** before commercial launch. According to **TechCrunch**, an earlier version of the order called for a review window of up to **ninety days**, but the **AI** industry pushed for a shorter timeline, closer to two weeks, which led to this thirty-day compromise (**TechCrunch**). ### An unexpected pivot toward more oversight This order marks a notable shift in direction for an administration that had until now favored an almost total deregulation approach to the **artificial intelligence** sector, actively dismantling several tech mandates inherited from the previous administration (**The Trends Wire**). The pivot toward a framework, even a voluntary one, for overseeing the most advanced models has surprised part of the industry. The **Washington Post** notes that the final version of the order limits the government's pre-release review period, an important nuance compared with earlier working drafts that some in the sector deemed too restrictive (**Washington Post**). *I readily admit I am surprised by this pivot from the **Trump** administration toward greater technological oversight. On national security grounds against rivals like China, I believe this added caution is justified, even though it contradicts the administration's usual deregulation rhetoric.* ## The companies directly affected by these new rules ### Anthropic and OpenAI already in the crosshairs According to **TechPolicy.Press**, export controls imposed in June already forced **Anthropic** to pull public access to its **Mythos 5** and **Fable 5** models, while **OpenAI** reportedly received a request to limit deployment of its new **GPT-5.6** model family to verified partners only (**TechPolicy.Press**). These concrete measures preceded the broader announcement of voluntary standards reported by the **Financial Times**. This targeted tightening around specific models suggests the American government is already, in practice, distinguishing certain levels of technological capability deemed advanced enough to warrant increased oversight, even within an officially voluntary framework. ### An industry divided on the need for these measures **Matt Pearl**, former director of emerging technologies at the **National Security Council** under the previous administration, told **Politico** that the current administration is engaged in an extremely difficult and complex balancing act between protecting national security and preserving innovation in **artificial intelligence** (**Politico**). He stressed the need for a clear, transparent, and fair process applicable to all model developers. This tension between national security and economic competitiveness runs through the entire current debate over regulating advanced **AI** in the **United States**, with no clear consensus yet emerging on the best way to resolve it. *I understand tech companies' frustration with shifting rules, but I also think a minimum of guardrails is necessary when we are talking about technologies as powerful as today's advanced AI models.* ## The risk of an unintended gift to China ### A concern raised directly by Politico A **Politico** article published on **July 1, 2026** argues that the **Trump** administration's hesitations and course changes on AI policy could, paradoxically, amount to a strategic gift to **China**, if American regulatory uncertainty slows domestic innovation while Beijing continues advancing without the same constraints (**Politico**). That is a serious criticism that deserves to be taken seriously. The reasoning is simple: if American companies must navigate a maze of shifting rules, variable review windows, and constantly evolving export requirements, they risk losing precious time in a technological race where every month counts against competitors less constrained by similar processes. ### Chip export controls toward Beijing It is worth noting that the **Trump** administration had previously authorized the sale of certain advanced **AI** chips to **China**, notably **Nvidia's** **H200** chips, while maintaining restrictions on the most advanced chips like the upcoming **Blackwell** and **Rubin** lines (**Politico**). This variable-geometry policy illustrates how hard it is to draw a clear line between commercial openness and strategic protection. I believe this question of controlling technology exports to strategic rivals like **China** will remain one of the most decisive files of the decade for the global technological balance of power. *On this specific point, I agree with **Politico's** criticism: it makes no sense to want to slow Chinese AI while selling certain advanced chips to Beijing. The West needs to pick a clear line and stick to it, not steer by sight based on whatever commercial pressure comes up.* ## The AI Action Plan, the underlying strategic roadmap ### A plan published as early as summer 2025 This new order builds on the **AI Action Plan** published by the **White House** back in **July 2025**, a strategic document calling on the United States to maintain its global dominance in **artificial intelligence** (**White House**). That plan had, among other things, tasked the **Department of Commerce**, through the **Center for AI Standards and Innovation** (**CAISI**) at **NIST**, with publishing assessments of advanced models coming out of the **People's Republic of China**. This strategic continuity shows that, despite the apparent tactical hesitations denounced by some observers, the administration maintains a coherent underlying line: staying a step ahead of its strategic rivals technologically, particularly **China**. ### No new federal regulatory agency planned A more recent document from March 2026, the **National Policy Framework for Artificial Intelligence**, states that the administration does not want to create a new federal regulatory agency for **AI**, preferring to rely on existing sector regulators and self-regulation standards developed by the industry itself (**White House**). This approach reflects a governance philosophy sharply different from the one adopted by the European Union on the same file. It is precisely this absence of a new regulatory structure that makes the voluntary model-review mechanism all the more significant: without a new dedicated agency, existing agencies will have to absorb this additional responsibility. *I remain torn on this American approach of largely letting the industry self-regulate. On one hand, I understand the desire not to smother innovation under bureaucracy; on the other, I worry a purely voluntary framework lacks teeth against security stakes this significant.* ## Cybersecurity, a blind spot turned priority ### An order first conceived as a digital shield According to an analysis by the **Benton Institute**, this order is primarily a cybersecurity directive aimed at strengthening federal government computer systems against AI-enabled threats, while also creating this voluntary framework for evaluating the most advanced models (**Benton Institute**). The text also directs federal prosecutors to pursue criminals who use **AI** to commit offenses. This defensive dimension of the order is often overshadowed in media coverage by the more spectacular questions tied to the race between major technological powers, even though it represents a substantial share of the actual content of the text signed on **June 2**. ### An information-sharing hub for software vulnerabilities The order calls for creating an AI cybersecurity information-sharing hub, in voluntary collaboration with industry and critical infrastructure operators, coordinating the detection and correction of software vulnerabilities across the country (**White House**). This initiative is meant, among other things, to protect sensitive infrastructure like rural hospitals, community banks, and local public services. It is a useful reminder that the stakes of **artificial intelligence** are not limited to the race for the most powerful models, but also touch very concretely on the resilience of everyday essential infrastructure. *I think this cybersecurity dimension deserves more public attention than it currently gets. Protecting a rural hospital from an AI-enabled cyberattack strikes me as just as urgent, if not more so, than winning the race for the most spectacular models.* ## What Congress is preparing on its own front ### A bipartisan legislative proposal According to **TechPolicy.Press**, Representatives **Jay Obernolte**, a Republican from **California**, and **Lori Trahan**, a Democrat from **Massachusetts**, published a **269-page** report on federal **artificial intelligence** policy, a sign Congress continues working in parallel with executive initiatives on this complex file (**TechPolicy.Press**). This bipartisan effort suggests there is still room for collaboration despite Washington's general polarization. The fact that such a bipartisan report emerged on a subject as sensitive as **artificial intelligence** reflects shared recognition, on both sides of the political aisle, of the strategic importance of properly regulating this technology without smothering it. ### A balance of power still to be found The coexistence of executive orders, voluntary frameworks negotiated directly with industry, and parallel legislative work in Congress illustrates well the institutional complexity the United States faces with a technology evolving faster than traditional regulatory processes (**TechPolicy.Press**). This multiplicity of actors can create confusion, but it also reflects a collective recognition of the file's urgency. Time will tell whether this fragmented approach eventually converges into a coherent framework, or whether it will keep generating the regulatory uncertainty that **Politico's** article on the risk of a gift to **China** rightly criticized. *I remain hopeful this bipartisan collaboration in Congress could produce something more stable than executive orders, which can change from one administration to the next. A law passed by Congress would carry more weight and predictability for the industry.* ## Reactions from Western allies ### Europe watching closely America's **European** partners are closely watching this US pivot toward a voluntary framework for advanced **AI** models, at a time when the **European Union** has already adopted a far more binding regulatory approach with its own legislative framework on **artificial intelligence** (**TechPolicy.Press**). This philosophical divergence across the Atlantic complicates transatlantic coordination on a file that is nonetheless crucial for collective Western security. Some allies fear that too fragmented an approach among Western allies could weaken the collective position against strategic rivals like **China**, which pursues a far more centralized and coordinated national technology strategy. ### Canada and its own AI questions **Canada**, a close ally of the **United States**, must also navigate this period of American regulatory uncertainty, as its own tech companies and **artificial intelligence** research centers rely heavily on access to American models and infrastructure (**TechPolicy.Press**). Any new American restriction on **AI** technology exports could have direct repercussions for the Canadian ecosystem. This North American interdependence makes it all the more important for **Ottawa** to closely follow this file's evolution, even though **Canada** has no direct seat at the negotiating tables between Washington and the major American tech companies. *As a columnist writing from a North American perspective, I believe **Canada** should have a more active voice in these discussions, even an informal one, rather than simply absorbing decisions made in Washington without being part of them.* ## Conclusion: between measured confidence and necessary vigilance ### A file I will keep following humbly This framework of voluntary standards for advanced **AI** models represents a sincere, if imperfect, attempt to reconcile technological innovation with national security amid intense competition with **China** (**Reuters**, **The Guardian**). I do not claim to know whether this approach will succeed, but I believe it deserves to be followed closely, with as much rigor as humility. The West, and the **United States** in particular, must maintain its technological edge over strategic rivals that do not apply the same ethical standards or the same security guardrails, while avoiding getting bogged down in bureaucracy that would hamper its own capacity for innovation. ### A vulnerability I fully own I will close this column by repeating what I said at the start: I am not a technical expert in **artificial intelligence**, and I will keep learning about this file in the months ahead, relying on credible journalistic and institutional sources rather than certainties I do not possess. *What I take away most from this file is that no administration, whatever its starting philosophy, can indefinitely ignore the security questions raised by technologies this powerful. The Trump administration's pivot toward more oversight, even voluntary oversight, is concrete proof of that.* **By Maxime Marquette, columnist** ## Columnist's transparency note ### Editorial stance I am not a journalist but a **columnist and analyst**. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies. I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. **I claim analytical clarity**, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events. ### Method and sources This text keeps the basic distinction between **verified facts and interpretive analysis**. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under **Sources**, never here. **Primary source families** used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies. **Secondary source families**: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations. When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under **Sources**. ### Nature of the analysis The analyses, interpretations and perspectives in the analytical sections of this article are a **critical, contextual synthesis** based on the information available, the trends observed and the expert comments cited in the sources consulted. My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors. This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text. ## Sources ### Primary sources - White House, order on advanced AI innovation and security — June 2, 2026 - Politico, Trump's AI policy reversals could benefit China — July 1, 2026 ### Secondary sources - Reuters via US News, Washington negotiates voluntary standards with AI companies — July 1, 2026 - The Guardian, Trump signs order for early access to new AI models — June 2, 2026 - TechCrunch, Trump signs narrower executive order on AI oversight — June 2, 2026 - TechPolicy.Press, monthly roundup of American tech policy — July 2, 2026 --- Source: Maxime Marquette, MadMax. When citing, name the author and link https://mad-max.co/en/article/j-ai-essaye-de-comprendre-le-nouveau-cadre-americain-sur-l-ia-de-pointe