The Download: Trump’s New AI Order, and Smart Glasses for Warfare

On a Tuesday that policy analysts and AI executives will likely parse for months, President Donald Trump signed a new executive order on artificial intelligence—his second attempt in as many weeks to shape the federal government's relationship with the most transformative technology of the decade. The order arrived less than 14 days after the White House scrapped its previous AI directive, and multiple sources describe it as a significant pivot in regulatory philosophy [1][2]. But the real story isn't just about what the order says—it's about what it doesn't say, and the $75 billion question of who ultimately controls the frontier of American AI development.

Meanwhile, on a parallel track that underscores the dual-use nature of modern technology, the same news cycle grapples with the emergence of smart glasses designed for warfare—a development that raises its own set of ethical and strategic questions about where augmented reality is heading when it leaves the consumer market behind [1]. Together, these two stories paint a picture of a technology sector caught between deregulatory impulses and national security imperatives, with the White House trying to find a middle ground that satisfies no one entirely.

The Architecture of the New Order: Voluntary Reviews and the Ghost of Regulation Past

The executive order signed Tuesday creates what The Verge describes as a "voluntary framework" for AI companies to share their frontier models with the federal government before public release [3]. The stated goal is to "promote secure innovation and strengthen the cybersecurity of critical infrastructure," a framing that carefully avoids the language of mandatory oversight that characterized earlier regulatory proposals [3]. This deliberately narrow document emerged in direct response to industry objections that torpedoed the previous iteration just weeks earlier [4].

TechCrunch's reporting confirms that the revised order emerged after significant pushback from major AI labs and their investors, who argued that mandatory prerelease reviews would slow innovation and cede competitive advantage to international rivals [4]. The White House appears to have listened: where the earlier order contained teeth, this one offers a handshake. Companies are invited—not required—to submit their frontier models for government review before deployment, a structure that critics argue amounts to little more than a suggestion box with presidential letterhead.

The timing is notable. Trump's first AI executive order was shelved less than two weeks ago, creating a vacuum of federal AI policy that the new order now attempts to fill [1]. The rapid turnaround suggests either a White House unusually responsive to industry feedback or one struggling to find a coherent AI strategy amid competing factions within the administration. Wired characterizes the process as one of fits and starts, noting that Trump "finally got on board" after initially shelving the original order [2]. The language of "finally" implies a certain reluctance, as if the president needed convincing that any form of AI oversight was politically or strategically necessary.

The order's text explicitly frames American AI success as a product of regulatory restraint, stating that the industry has prospered "because we refuse to stifle this innovation with overly burdensome" requirements [3]. This directly echoes the industry's own talking points, signaling that the White House has fully internalized the argument that American AI dominance depends on keeping government at arm's length. Whether that assumption holds true in an era of increasingly capable and potentially dangerous models remains an open question—one the order deliberately leaves unanswered.

The Numbers Behind the Narrative: $75 Billion and the Stakes of Voluntary Compliance

The MIT Technology Review's coverage drops a cluster of data points that deserve careful examination: the order references $75 billion in projected AI investment, along with figures of 50% and 60% that appear to relate to market growth or capability benchmarks [1]. While the sources do not specify the exact referents for these percentages, the scale of the numbers underscores the economic weight of the decisions being made. This is not a niche technology sector; it is an industry on track to reshape the global economy, and the White House is essentially asking companies to voluntarily submit their most valuable intellectual property for government inspection.

The voluntary nature of the framework creates a classic collective action problem. Companies that submit their models for review gain a government seal of approval but also risk exposing proprietary architecture and training methodologies. Companies that decline to participate avoid that risk but may face public pressure or future regulatory consequences if their models cause harm. The order does not specify what happens to companies that opt out, leaving a significant ambiguity that industry lawyers will likely spend the coming weeks parsing.

This is where the divergence between sources becomes instructive. The Verge emphasizes the cybersecurity and critical infrastructure angle, framing the order as a national security measure [3]. TechCrunch focuses on the industry objections that shaped the final document, presenting the order as a negotiated settlement between the White House and tech giants [4]. Wired takes a more process-oriented view, highlighting the political maneuvering that led to the signing [2]. The MIT Technology Review, as the original source, provides the broadest framing, connecting the AI order to the parallel story of military smart glasses [1]. Each source captures a different facet of the same event, and together they reveal an administration still figuring out its AI posture in real time.

Smart Glasses for Warfare: When Augmented Reality Goes Tactical

The second thread running through this news cycle is the development of smart glasses for military applications—a story that the MIT Technology Review pairs with the AI order in a single edition of its Download newsletter [1]. This is not an accidental juxtaposition. The same technologies that power consumer augmented reality—computer vision, real-time data overlay, spatial computing—are being adapted for battlefield use, raising questions about the militarization of technologies originally developed for entertainment, productivity, and social connection.

Smart glasses, as defined by the broader technology literature, are "eye or head-worn wearable computers" that can include displays adding information alongside or to what the wearer sees, or alternatively, glasses that can change their optical properties electronically [1]. The military applications of this technology are extensive: heads-up displays for soldiers that show enemy positions, tactical data, and biometric information; augmented reality systems for drone operators that overlay flight paths and target data; and training systems that simulate combat environments with unprecedented fidelity.

The convergence of AI and smart glasses is where the story gets particularly interesting. The AI executive order's focus on frontier models and cybersecurity has direct implications for military smart glasses, which rely on sophisticated AI systems for object recognition, threat assessment, and real-time decision support. If the government reviews AI models voluntarily, what happens when those models are embedded in weapons systems? The order does not address this question directly, but the logical extension of its framework suggests that military AI systems fall under a different regulatory regime entirely—one that is classified, opaque, and operating on timelines that bear little resemblance to the commercial sector.

Winners, Losers, and the Developer Friction Problem

The immediate winners of the new executive order are the large AI labs that lobbied against mandatory review. Companies like OpenAI, Anthropic, Google DeepMind, and Meta's AI research division can now point to the voluntary framework as evidence that the government trusts them to self-regulate—a powerful narrative tool in both public relations and future regulatory battles. The losers are the safety advocates and civil society organizations that pushed for mandatory prerelease testing of frontier models, particularly those with capabilities that could pose existential or catastrophic risks.

But a third category deserves attention: the developers and startups building on top of foundation models. For these companies, the voluntary framework creates uncertainty. If a major model provider submits its latest release for government review, the review process could delay access to the model, creating competitive disadvantages for startups that depend on timely access to frontier capabilities. If the provider declines to submit, the startup faces the opposite problem: deploying a model that lacks government validation, potentially exposing the startup to liability if something goes wrong.

The sources do not specify the timeline or mechanics of the review process, creating a vacuum of information that will likely be filled by speculation and legal interpretation. This ambiguity is itself a form of friction—developers cannot plan around a process whose parameters are undefined. The order's language about "promoting secure innovation" suggests that the White House is aware of this tension, but the voluntary structure leaves the hard questions to be resolved later, presumably through additional rulemaking or industry self-governance.

The Macro Trend: Deregulation as Industrial Policy

Stepping back from the specifics of the order, a broader pattern emerges. The Trump administration's approach to AI regulation mirrors its approach to other technology sectors: prioritize American competitiveness, minimize government intervention, and trust market forces to manage risks. This is not a neutral policy choice; it is an explicit industrial strategy that bets on speed over caution, private sector judgment over public oversight, and competitive advantage over precautionary principles.

The $75 billion figure referenced in the MIT Technology Review's coverage provides a useful anchor for understanding the stakes [1]. That level of investment creates powerful constituencies with aligned interests: venture capitalists who need exits, founders who need regulatory clarity, and employees who need their companies to succeed. The voluntary framework satisfies these constituencies while providing enough rhetorical cover to claim that the government is taking AI safety seriously. Whether it actually improves safety outcomes is a separate question—one the order's structure seems designed to avoid answering.

The smart glasses for warfare story adds a darker dimension to this analysis. If the commercial AI sector receives wide latitude to develop and deploy frontier models, and if those same technologies simultaneously adapt for military use, the line between civilian and military AI becomes increasingly blurred. The voluntary review framework applies to commercial models, but what about dual-use technologies that start in the private sector and end up in defense applications? The order does not address this pipeline, leaving a regulatory gap that could have significant consequences as AI capabilities continue to accelerate.

What the Mainstream Media Is Missing

The coverage of the executive order has focused heavily on the procedural drama—the scrapping of the previous order, the industry pushback, the signing ceremony. What has received less attention is the fundamental question of whether voluntary review can work at all for frontier AI models. The history of technology regulation suggests that voluntary frameworks tend to work well when the risks are low and the incentives for compliance align with business interests. They tend to fail when the risks are high and the incentives point in the opposite direction.

Frontier AI models present exactly the kind of high-risk, misaligned-incentive scenario that voluntary frameworks struggle to address. A company that discovers a dangerous capability in its latest model faces a choice: disclose the capability and delay deployment, potentially losing market share to competitors, or keep quiet and release the model, hoping that nothing goes wrong. The voluntary framework provides no mechanism to make the first option more attractive than the second. This is not a failure of the order's design; it is a feature of the deregulatory philosophy that underpins it.

The smart glasses story compounds this concern. Military applications of augmented reality and AI are advancing rapidly, often outside the public eye and without the transparency that characterizes civilian AI development. The convergence of these two trends—deregulated commercial AI and opaque military AI—creates a landscape where the most dangerous applications of the technology are the least visible and the least accountable. The MIT Technology Review's decision to pair these stories in a single newsletter suggests an editorial awareness of this convergence, even if the individual articles do not explicitly connect the dots.

The Road Ahead: Ambiguity as Strategy

For now, the AI industry has gotten what it wanted: a light-touch regulatory framework that preserves flexibility and avoids mandatory requirements. But the order's ambiguity cuts both ways. The same vagueness that allows companies to interpret the voluntary framework favorably also allows future administrations to reinterpret it more strictly. An executive order is not a law; it can be modified or revoked by the next president, and the current order's reliance on voluntary compliance makes it particularly vulnerable to political shifts.

The $75 billion investment figure suggests that the industry is betting on continued regulatory favorability [1]. If that bet proves wrong—if a major AI incident occurs, if public opinion shifts, or if a new administration takes a harder line—the voluntary framework could become the foundation for mandatory requirements rather than an alternative to them. Companies that participate in the review process now are not just complying with a request; they are creating precedents and establishing norms that could become the baseline for future regulation.

The smart glasses for warfare story serves as a reminder that AI regulation is never just about commercial products. The same technologies that power chatbots and image generators are being integrated into weapons systems, surveillance infrastructure, and military command-and-control networks. The executive order's focus on "critical infrastructure" acknowledges this reality, but its voluntary structure leaves the hardest questions unanswered. As the lines between civilian and military AI continue to blur, the White House's bet on voluntary compliance will face its most serious test not in the boardrooms of Silicon Valley, but on battlefields where the consequences of AI failure are measured in lives rather than market share.

---

References

[1] Editorial_board — Original article — https://www.technologyreview.com/2026/06/03/1138322/the-download-trump-ai-order-smart-glasses-warfare/

[2] Wired — This Is How Trump Finally Signed the AI Executive Order — https://www.wired.com/story/this-is-how-trump-finally-signed-the-ai-executive-order/

[3] The Verge — Trump signs executive order to review AI models before they’re released — https://www.theverge.com/policy/941775/trump-ai-executive-order

[4] TechCrunch — Trump signs narrower executive order on AI oversight after industry objections — https://techcrunch.com/2026/06/02/trump-signs-narrower-executive-order-on-ai-oversight-after-industry-objections/