The same censorship apparatus the Party built to control its people is now corrupting the AI systems its leaders depend on. The United States, by leaning into an open marketplace of information and ideas, will gain advantage as it takes a different path.

AI is increasingly training newer, faster AI models. This typically involves scraping the internet for content and then loading it into datasets for new programs. The problem: online content used for training is increasingly generated by AI. As a result, each generation of technology drifts from reality.

AI researchers refer to this as “model collapse,” a phenomenon in which models trained on their own synthetic outputs degrade over successive generations. The only defense is a constant influx of fresh, honest, human-generated information. Without it, the system folds in on itself.

China’s Great Firewall cuts off that influx, expediting the impact of model collapse within its borders. The first generation of large language models worldwide was trained on massive datasets of publicly available human-generated text. These human-derived pieces of information built an algorithmic approximation of how people think, argue, explain, and communicate.

Now, the internet is filling with AI-generated content at a rate inconceivable five years ago. Marketing copy, product descriptions, social media captions and news summaries are increasingly produced by AI systems and published online. They’ve all become generic and detached from human signals.

With each cycle of new AI, the models drift further from their human origins. The newer systems amplify the patterns AI systems favor while losing nuance and amplifying existing biases. Each successive AI generation is one step further removed from the humans these programs were originally built to serve.

In China, the Great Firewall accelerates this problem. The Chinese Ministry of Public Security built the Great Firewall of China in the late 1990s to censor the Chinese people. It is now the most sophisticated information control infrastructure in human history. The Great Firewall does not just restrict what Chinese users can see. It shapes the data used to train Chinese AI systems. By design, it strips out politically sensitive events, dissenting viewpoints, and independent reporting. Left behind is a curated record of reality aligned with the Party’s narrative. That filtered record of information becomes the raw material for LLMs.

The training data fed into LLMs in China does not contain any criticism of the government, fair reportage of controversial topics, or accurate information about Chinese history. Events such as the the Uyghur detention camps exist inside the Firewall only as the state chose to describe them.

Now add model collapse on top of this. Chinese AI companies such as Baidu, Alibaba, ByteDance and dozens of others aggressively deploy AI-generated content across their platforms. This material becomes training data for the next generation of Chinese AI models. With independent reporting locked out, model collapse expedites inside the Great Firewall with no escape valve. The practical consequences of this divergence are already visible and will accelerate.

Chinese LLMs struggle with tasks that require new observation, original synthesis or human complexity that the training data was designed to suppress. If asked about repression in Chinese history or controversial current events such as the Uyghur detentions, these LLMs either do not answer or produce a response indistinguishable from a Party press release. A Chinese trade official relying on domestic AI to model the economic impact of Western sanctions is working from a system incapable of providing an honest account of how those sanctions functioned or failed in comparable historical cases.

The West has a milder version of this problem.

Western AI models are trained on increasingly synthetic content, but human reporters regularly push new information into the ecosystem. Free and open societies have structural advantages to retain the capacity to reason about what is happening in the world rather than what previous AI systems described. When asked about Tiananmen Square, Western accounts typically focus on the 1989 protests and the crackdown. By contrast, Chinese models either refuse to answer or return state-aligned language. This information gap becomes part of the training data for the next generation.

To maintain a competitive AI advantage over China, Washington should treat human-generated data as a strategic asset and invest in journalism, open web archives, and synthetic-content labeling. Preserving the integrity of American training data is a defense imperative, not a tech problem.

Chinese leaders deploying AI products to make decisions about economics, geopolitics, and public health will make those decisions based on systems trained on what China’s information control apparatus wants people to believe. That is not an intelligence system. It is a mirror. And the tragedy of model collapse is that a mirror that has been looking at itself long enough no longer reflects anything.

Joe Buccino is a retired U.S. Army colonel and the author of “When Every Word Counts: How to Earn Trust, Command Attention, and Communicate Clearly in Any Situation.”

It is an early glimpse of a broader problem that many organizations would prefer not to acknowledge: As AI is rushed into critical systems, it is introducing new failure modes faster than they can understand or control them.

For defense organizations increasingly integrating AI into mission-critical systems, the implications are far more consequential.

When organizations pause to consider these risks at all, they often reach for a familiar reassurance: there will be a “human in the loop.” The idea is that even if the system is complex or unreliable, a person will catch mistakes before they matter.

This reassurance is dangerously misleading. A “human in the loop” whose sole function is to approve a machine’s actions is not a safeguard but a design failure. Attention wanes because nobody can concentrate on a job that is mostly doing nothing, and over time the operator’s skills atrophy to the point that they cannot meaningfully supervise the system. What remains is the appearance of oversight rather than the reality.

In military contexts, this kind of degraded human involvement is not just inefficient but operationally dangerous.

This pattern is not new. Engineers have seen it before, most famously in the Therac-25, a radiation therapy machine introduced in 1982. It combined the functions of two predecessor systems in a smaller, more convenient package, and its improved automation made it faster and easier to operate. Safety was “guaranteed” by the presence of a human operator who had to confirm actions – in effect, a “human in the loop.”

The system failed anyway. Patients began developing severe radiation burns. Hospitals dismissed the possibility of machine error, and the manufacturer insisted overdoses were impossible. Only after sustained investigation was it discovered that the machine contained multiple safety-critical software flaws. By then, six overdose accidents had occurred, three of them fatal.

The deeper problem was not just faulty code but faulty design. The machine frequently halted with poorly explained error messages, requiring operators to “press P to proceed” to continue treatment. Because these errors were common and rarely meaningful, operators became habituated to restarting the system dozens or hundreds of times a day. When real malfunctions occurred, the act of “operator confirmation” had already lost its meaning. In one case, an operator restarted the machine multiple times, unknowingly delivering repeated overdoses. The presence of a human operator did not prevent the failure; it normalized it.

Today, we are repeating this mistake. Computer scientists are rushing to incorporate poorly understood AI systems into safety-critical environments, and when concerns are raised they are often waved away with the same phrase: there will be a human in the loop. This assumption is now appearing in discussions of defense systems, from decision support to autonomous operations.

People will argue that AI is fundamentally different, and in one sense they are right. We have never before deployed systems whose behavior is explicitly probabilistic and nondeterministic in high-stakes environments. In defense contexts, where uncertainty compounds quickly and errors can cascade across systems, this is especially concerning. But AI is also not different in the ways that matter most. It is still software, embedded in larger systems composed of people, processes, and machines. It cannot act in the real world without that surrounding system, and those systems fail in ways that are already well understood. Engineers and operators have spent decades studying how complex, tightly coupled systems behave under pressure.

What we are seeing now is not a new class of failure but a familiar one, accelerated. The software industry is once again demonstrating an inability to learn from its own history. That would be unfortunate if we were only talking about Spotify recommendation algorithms. It becomes dangerous when these same patterns are introduced into the systems that organizations — and nations — depend on.

Recent Pentagon leaks suggest that AI systems may already be influencing where bombs land. In such environments, the illusion of human oversight is worse than no oversight at all. It creates confidence without control.

If we spend the next decade hiding unsafe systems behind the fig leaf of the “human in the loop,” the consequences will not be theoretical.

Mikey Dickerson was the founding administrator of the U.S. Digital Service and is a crisis engineer at Layer Aleph. He is a co-author of the forthcoming book Crisis Engineering.

Armored vehicles are quickly spotted and destroyed with either drones or artillery. Soldiers on foot seldom fare any better. Negating the other side’s drone capabilities would be a tremendous advantage, but conventional air defense isn’t good enough.

Fortunately, the U.S. has developed a solution: 30mm chain guns — traditionally mounted on Apache attack helicopters — bolted to civilian pickup trucks and connected to a portable sensor called Mobile–Acquisition, Cueing and Effector, or M-ACE.

After detecting drones, the Northrop Grumman-made system calibrates programmable shells to detonate mid-air, meaning the system, which is cost-friendly compared to other solutions, can destroy quadcopters and dismantle swarms without hitting them directly.

Such an answer to some of the evolving drone challenges seems too good to be true, but its viability has been vetted in recent years.

Ukraine has already deployed M-ACE on the battlefield in small quantities, along with Australian “Slinger” systems, which use the same chain gun. Taiwan has also expressed interest in the system, iterations of which have recently been fielded by the U.S. Army and Marine Corps.

A key shortcoming with these systems, however, is driving nonmilitary vehicles in war zones. While the mounted gun may be able to stop drones, the vehicle’s crew would have effectively no protection from landmines, artillery or gunfire.

To make these counter-drone platforms truly effective, systems should be mounted on remotely operated vehicles. Enter the Tesla Cybertruck.

The Cybertruck’s usefulness for Ukraine cannot be unpacked without acknowledging its failure on the civilian market. Elon Musk promised hundreds of thousands of sales each year; Tesla fell short by 92% in 2025, barely selling 20,000 of the widely mocked pickups.

Consumers have a slew of reasons to avoid Cybertrucks. In addition to political controversy surrounding Tesla’s CEO, the truck’s design has been slammed for poor visibility, accelerator pedals that get stuck, warranties voided by car washes, dysfunctional windshield wipers and trunk doors that have gone viral for safety concerns.

Such issues, however, would be mostly irrelevant should the vehicle be used to aid the Ukrainian military.

Low demand, meanwhile, has left over 10,000 Cybertrucks sitting unsold in dealership lots. With Tesla modifying its “Gigafactory” in Austin to build a quarter-million units each year, Cybertrucks could, in theory, be delivered quickly to Ukraine — and in large quantities.

Availability isn’t the only factor that makes Tesla’s pickup a good choice. With systems like M-ACE limited by the vulnerability of crew members, Tesla could bypass the supervision requirement of the vehicle’s self-driving capability.

With remote operation, Cybertrucks on the battlefield would add a layer of safety other M-ACE-equipped pickups can’t provide. These vehicles are also easier to mass produce than any purpose-built unmanned ground vehicle rated to handle 30mm autocannons.

Today, two UGVs fit this role — the Estonian-made THeMIS and the U.S.-made Textron Ripsaw M5. Per-unit prices are not publicly available, but it’s safe to assume that these systems cost hundreds of thousands: THeMIS has sold for several million dollars, and the Ripsaw M5’s civilian variant has a starting price of $295,000.

If the Cybertruck is a viable substitute for these UGVs, an $80,000 price tag seems like a bargain. Beyond its availability and self-driving mode, the Cybertruck also offers the benefit of electrical power.

EVs have several logistical and functional advantages over fuel-powered trucks. The lack of moving parts and fuel requirements make them easier to maintain and cheaper to operate. They also create less noise and less heat, key for avoiding attention from Russian forces — especially those with thermal cameras.

A fleet of Cybertrucks equipped with chain guns and sensors to clear the skies of small drones could significantly affect the war and tilt certain battlefields in Ukraine’s favor.

Defanging enemy drones would, at worst, make the war effort more sustainable by reducing casualties and buying time for Europe to supplement Ukraine’s efforts.

More optimistically, these systems could give Ukraine’s ground forces more flexibility to move on drone-saturated front lines.

There’s no sign that an end to the war, or an improvement in Cybertruck sales, will come any time soon. Ukraine could be the best backup plan for Tesla’s misplaced investment.

Alex Lee is a recent graduate from Washington University in St. Louis, where he majored in political science and minored in writing.

Russia’s full-scale invasion of Ukraine has accelerated the return of high-intensity competition to the High North, a region that was perceived as a peripheral corner of global geopolitics for much of the post–Cold War era. As the United States and NATO Arctic allies hold intense debate about security throughout the region and begin to integrate drones for collective defense and deterrence, they face an uncomfortable reality: Russia is moving faster.

A growing gap has emerged between the alliance’s ambition to defend the High North and its ability to field Arctic-capable uncrewed systems at scale. The future of Arctic security will hinge not only on submarines, missiles, fighter jets and icebreakers, but on the capacity to deploy, sustain and counter uncrewed systems in extreme conditions and unprecedented numbers.

In this specific competition, Russia can leverage a significant edge. It fields the world’s largest industrial-scale drone ecosystem outside China and is institutionalizing combat lessons from Ukraine. Like Kyiv, Moscow has established a dedicated branch for uncrewed systems, expanded mass-training of drone operators, and is forming new drone units across the joint force, including within the Northern Fleet. It is also investing in Arctic-adapted platforms and command centers for long-range maritime drones. Annual production now exceeds 1.5 million units, with Western intelligence services expecting sharp increases driven by Chinese industrial support and sanctions evasion.

In the Arctic, Russia is likely to use drones as integrated force multipliers: enabling persistent ISR along Arctic littorals and the Northern Sea Route, cueing Bastion coastal defense systems, supporting anti-submarine and anti-surface operations and conducting long-range strikes with one-way attack drones alongside traditional missiles. These activities will be tightly coupled with aggressive cyber and electronic warfare to degrade allied sensing, targeting and command-and-control capabilities.

For NATO, uncrewed systems may matter more in the Arctic than in any other theater. Vast distances, sparse infrastructure and extreme weather constrain human presence and amplify gaps in domain awareness and response times. Traditional crewed platforms such as maritime patrol aircraft, submarines and ice-capable vessels remain essential but are limited in number and costly to sustain. Even with renewed investments, they cannot alone provide the persistent surveillance, logistics, and targeting required by NATO’s regional defense plans.

Drones help fill this gap. They can offer unmatched persistence, scalability, and flexibility at lower operational cost and risk to personnel, conducting a wide array of tasks, from reconnaissance and surveillance to logistic and resupply, casualty evacuation, mine and countermine operations, strike and more. In short, uncrewed systems are no longer niche enablers but indispensable elements for strengthening deterrence and defense in the Arctic.

Technology does not work in a vacuum

Procuring advanced platforms, however, does not automatically translate into usable capability. NATO faces structural challenges that limit its ability to exploit drones in the High North.

First, most systems are not designed for Arctic conditions. Extreme cold degrades batteries, icing affects propulsion and sensors, communications become unreliable, and corrosion accelerates wear. As a result, NATO suffers from a shortage of uncrewed systems certified for sustained cold-weather operations.

Second, infrastructure is sparse, complicating logistical support. Airfields, ports, and repair facilities are limited, while satellite coverage above 75°N is degraded and less reliable. Therefore, any robotization strategy must include adequate sustainment across the board. These challenges make reliance on autonomous features and GNSS-denied navigation essential, though the latter raises system complexity and costs.

Third, NATO’s problems are not only technical but also institutional and organizational. Doctrine, training, and innovation pathways have not kept pace with the rapid evolution of uncrewed technology and its military use.

Doctrine, personnel and other “unsexy” things

Despite recent updates, NATO doctrine still treats drones largely as supplements rather than as core elements of deterrence and defense. Commanders lack mature concepts for integrating uncrewed systems into multidomain operations against a drone-enabled adversary like Russia. The Arctic environment magnifies these gaps and demands greater efforts to develop shared approaches to human–machine teaming, standardized tactics, deployable sustainment, and counter-uncrewed systems operations.

Recent initiatives to support concept development, standardized training and operational experimentation are promising but insufficient.

Personnel challenges compound the problem. With many NATO allies struggling to recruit and retain people, the skilled operators, software specialists and specialized maintainers required for Arctic drone operations are in high demand but few in numbers.

Russia, by contrast, is rapidly expanding its training pipelines and embedding uncrewed expertise across its force structure. Autonomy can mitigate but not fully resolve the personnel problem. The human element and doctrine remain decisive to leverage the technological edge that uncrewed systems provide.

Innovation pathways across the alliance also remain slow and fragmented. While NATO experimentation initiatives are growing, they struggle to scale results. At the same time, private capital is beginning to play a larger role, with venture capital firms and European public–private mechanisms such as the European Defence Bank and European Investment Fund (EIF) increasingly backing uncrewed systems and AI startups.

However, as we explain in a recent report, harnessing this momentum will also require procurement reforms that favor speed, scale, and interoperability over risk-averse procurement programs and industrial protectionism.

A strategic imperative

For NATO as well as Arctic allies, the central question is whether they can absorb the institutional, organizational, and operational change required by large-scale robotization. The priorities are clear: field Arctic-ready uncrewed systems that are modular and interoperable by design; connect the battlespace through resilient command and control structures and high-bandwidth data pipelines; devise robust personnel and sustainment strategies and develop doctrinal tenets and concepts to effectively integrate uncrewed systems into the force structure and multidomain joint operations in an austere environment; and procure collectively and at scale, including through a mix of national and multilateral initiatives.

Without a strong demand signal for key platform categories such as tactical ISR drones, loitering munitions and uncrewed logistics vehicles, allies will continue paying premium prices for bespoke systems while Russia floods the region with battle-proven, mass-produced platforms.

These steps demand resources and urgency but also a solid understanding of what drones can and cannot do. Decision-makers need to move beyond the misleading quantity versus quality debate and focus on both elements, for uncrewed systems are neither silver bullets nor replacements for traditional lethality. Instead, their value lies in complementarity.

Effective Arctic defense requires a balanced high–low mix in which uncrewed systems extend and multiply the effectiveness of traditional forces rather than substitute for them.

Federico Borsari is a fellow with the Transatlantic Defense and Security program at the Center for European Policy Analysis (CEPA). He focuses on defense technology and military affairs.

Gordon B. “Skip” Davis Jr is a senior fellow with the Center for European Policy Analysis (CEPA), a retired U.S. Army major general, and former NATO deputy assistant secretary general for defense investment.

On Jan. 1, an order went into effect providing Russian companies with expanded income tax incentives for buying Russian unmanned aerial vehicles. For every ruble they spend on UAVs, those companies can deduct two rubles from their income taxes. On its face, it’s a pretty good deal, especially for those companies hoping to contribute to the state’s war machine.

But the reality, including what this means for Ukraine and the West, is more complicated.

Russia’s decade-plus domestic tech push shows that Russia can excel at software development but often fails miserably at hardware, yielding persistent efforts to steal Western technology components and growing dependence on buying Chinese ones. Even UAV-focused tax incentives are unlikely to fix that dynamic. At the same time, pushing Russian firms towards Russia’s UAV sector could help prop up those companies, supplying the cash to keep charging ahead. And it could ensure even fewer Russian firms can commercialize outside the security state’s reach —giving Moscow further leverage over technologies it can weaponize against the West.

Over the last decade, Russia has accelerated its push to promote a domestic technology sphere —one driving towards autarky and securitization. This mixed-success effort has sought to expel Western technology (seen as a foreign espionage threat) and build up Russia’s technology sectors. Since February 2022, Western sanctions have only catalyzed Russia’s technological isolation, both encouraging and forcing it to double down on domestic technology to mitigate its crumbling economy and keep fueling its war machine. This led to the creation of the high-tech tax incentive list in July 2023.

Enter this latest update to boost the Russian drone industry. It’s no secret that Russian UAVs have played a central role in the Putin regime’s full-scale, illegal war on Ukraine, and Ukraine’s defense against it, over the past four years. A January report from one of Latvia’s intelligence agencies estimated that drones are responsible for 70-80% of injuries and deaths on both sides of the war in Ukraine. Changes are rapid: Sam Bendett, a leading expert on Russian and Ukrainian UAVs, has assessed that Ukrainians and Russians are developing functionally distinct versions of UAVs and UAV defenses on innovation cycles ranging from two weeks to three months. Now, some battlefield UAVs are increasingly using AI models to chase and strike targets.

But where the components in Russia’s drones come from spells out precisely why Russia’s new policy may make little difference to Ukraine’s war effort and those in the West supporting it.

Not all Russian-used UAVs have entirely been, in components or design, Russian. The Kremlin’s security apparatus has smuggled countless microelectronics and other components from the West into the country’s military systems and machinery. Russia continues to buy more and more technology from China to funnel into its defense and technology base, including to support its UAVs. Even Russia’s most recent UAV battlefield innovations allegedly contain European, Chinese, and apparently US-made components — hardly symbols of Russian technological autonomy. Meanwhile, Iran remains a major global producer of attack UAVs and has played a significant role, both as a supplier and a teacher, in helping Russia bolster its drone manufacturing. Countries Russia could buy UAVs or UAV parts from, such as China and Turkey, have replicated Iran’s cheap, simple Shahed-136 design, too.

These trends are corroborated by a key lesson from Russia’s decade-plus domestic technology effort: Russia can excel in many ways at software development, such as building an alternative operating system to Microsoft Windows; however, it has failed time and time again, sometimes with embarrassing spectacle, to stand up domestic hardware manufacturing without significant informational or technological help from the outside. For the West, the tax incentive for UAV purchases may therefore be a better signal of the Kremlin’s intent — keep leveraging drones in all-out war on Ukraine — than of long-term self-sustainability. Hence, Ukraine, the European Union, and other players can keep putting pressure on Russia’s illicit procurement networks and on the other governments, including China and Iran, enabling its UAV innovations.

Simultaneously, the Russian government pushing Russian firms towards Russia’s UAV sector could help supply those companies with funds they will not get from, say, participating in the Western arms market or organically spreading into civilian marketplaces. Those additional revenues for large and small UAV developers alike could be meaningful, giving them the capital they need under early-day or stressful innovation cycles to keep retaining talent and (probably illicitly) acquiring the needed non-Russian tech. In response, the West may find itself trying to undercut Russian UAV efforts that hinge more on Russian domestic incentives further out of their reach. Limiting sales of Western UAV components to known Russian fronts, for example, is at least more in the power of Washington, London, Berlin, or Kyiv than trying to use sanctions and other tools to reach within Russia and shift internal market dynamics on wartime UAVs.

But perhaps the biggest strategic implication for the West may be the latest tightening of the Putin regime’s grip over technology in Russia. The Kremlin has spent decades continually bolstering its ability to capture and coerce domestic technology actors to do its bidding—or to at least stay in line. Especially since February 2022, however, it has become increasingly difficult for companies to seriously commercialize in Russia without turning to the military-industrial complex. That’s bad news for the West. Ventures whose capital needs are high up front are more likely to turn to the defense base for funding. Companies looking to sell internationally know they will need to not just bend the knee if asked but actively court Russian defense contracts.

And even though a tax incentive didn’t cause many Russian UAV firms to eagerly support the Putin regime’s slaughter in Ukraine, it certainly won’t encourage them to stop it.

Justin Sherman is the founder and CEO of Global Cyber Strategies, a DC research and advisory firm, an adjunct professor at Georgetown University, and the author of “Navigating Technology and National Security.”

The Trump administration has a rare chance to change all that. Last night, the administration announced that it would make widespread changes to its interpretation of the MTCR as part of a broader effort to reform the way the U.S. sells and transfers weapons abroad. This step would face up to today’s realities and help Washington compete far better in today’s increasingly crowded arms market, though, as of yet, there is no clarity on whether the reinterpretation of the MTCR would be matched with clear principles on which UAVs will be acceptable to export and to whom.

Efforts to revise the U.S. approach to the MTCR date back to the first Trump administration, which in 2020 announced that it would reinterpret the regime to treat certain UAVs as Category II technologies instead of the more restrictive Category I, which historically carried a presumption of export denial for UAVs. This effort endured in the Biden administration, which in 2025 announced a revised interpretation of the MTCR to allow more flexible interpretations of Category I, but explicitly excluded the potential for transfers of production facilities and technologies. The second Trump term has picked up right where it left off: The White House issued an executive order in April to streamline defense exports, which was quickly followed by Congressional action, including H.R. 3068 and the Streamlining Foreign Military Sales Act of 2025.

Taken together, the coordinated action between the executive and legislative branches represents a concerted effort to make some of the most comprehensive changes to arms sales and transfers in decades. Nested under this effort is an understanding — shared across party lines — that the competition with China is increasingly expanding to the international arms market. China is a top-five global arms exporter, and has its sights set on expanding its position. Beijing recently displayed new armaments to dozens of foreign heads of state at its military parade, and Chinese officials have been seeking to capitalize elsewhere on the performance of their weaponry in the Pakistan-India crisis earlier this year. Increasing the speed and efficiency of U.S. arms transfers is one way to preclude China from making inroads in new export markets.

UAV sales are increasingly at the forefront of this competition. Sales of armed drones have ticked up dramatically in recent years as countries have witnessed the benefits of utilizing uncrewed, relatively cheap technologies in combat. A new MTCR interpretation opens up the aperture of American UAVs available to countries around the world. U.S. UAVs like the MQ-9 Reaper are likely to be in top demand, as Washington has historically limited the amount of external operators of this advanced platform to select NATO allies and Major Non-NATO Allies (MNNAs). For the United States, expanding sales of MQ-9s would have the dual effect of increasing interoperability with more partners while also blocking out China and others from exploiting a gap in the market.

Nowhere does this competition play out more acutely than in the Middle East. Years of asymmetric warfare and swarm tactics have only increased the demand for large, cheap, uncrewed capabilities. Faced with repeated rejections from Washington, many have turned to outside suppliers, like China. Chinese drones have already been sold to traditional U.S. partners in the region including Jordan, Saudi Arabia, the UAE, Egypt and Iraq. Some have even coordinated with Beijing on manufacturing UAVs. With a new MTCR interpretation in place, the Trump administration seems poised to reverse course and begin exporting UAVs to the region. The U.S. approved the sale of 8 MQ-9Bs to Qatar in March, and Saudi Arabia is reportedly seeking up to 100 MQ-9s as part of a massive arms deal negotiated during Trump’s visit in May.

Revising the U.S. approach to the MTCR is long overdue. As competition with China seeps into other arenas, maintaining a constellation of interoperable allies and partners is crucial for the U.S. Doing so requires meeting market demand, and the demand worldwide — after witnessing conflicts in Europe and the Middle East — is for UAVs. Crucial for the U.S., however, is finding a way to balance meeting this demand with protecting its proprietary information. Beyond the historically strict interpretation of the MTCR, American officials have been reticent to export advanced UAVs for fear that their technologies and capabilities would be vulnerable to adversary collection. This concern is warranted, and U.S. officials should couple the roll-out of this reinterpretation alongside the establishment of strict nonproliferation parameters with partner countries and a robust oversight mechanism to ensure American technologies are not in jeopardy and countries or non-state actors do not take advantage of more lax policy.

Ultimately, for the United States to remain the partner of choice, not just in the Middle East but worldwide, it must continue to balance between meeting market demand and protecting its proprietary technologies. But the balance must also reckon with the original purpose of the MTCR: Preventing the proliferation of missile technologies that could destabilize regions or fuel arms races. Providing the U.S. the ability to simply compete with China in UAV sales is not, in itself, a strategy. The reinterpretation should be coupled with updated nonproliferation standards and robust safeguards to balance both competition and global responsibility. Balancing both can lead to deeper, more sustained defense collaborations in the Middle East and around the world.

Elizabeth Dent is the Nathan and Esther K. Wagner Senior Fellow at the Washington Institute, where Grant Rumley is the Meisel-Goldberger Senior Fellow. Both previously served in the Office of the Secretary of Defense.

Fast forward to today and a similar tension has formed in the field of defense wargaming, where traditional wargamers — relying on years of expertise and bespoke game designs — are coming to grips with rapid advances in modeling and simulation and artificial intelligence.

At the Johns Hopkins Applied Physics Laboratory, we have been living and breathing that tension as we have worked to incorporate generative AI and modeling and simulation into defense wargaming. The results of that work? We don’t think we need a 20-year debate. Just like in baseball, the future of wargaming lies in a marriage of modeling and simulation, human expertise and AI.

To understand why wargaming is having its “Moneyball” moment, you have to first unpack what makes traditional wargaming so valuable. Wargaming is fundamentally about human decision-making, but its magic is in the experiential learning opportunities the games provide. War is never simple. There is no “all-seeing eye” that provides perfect information. Hence, wargaming explores how humans make decisions in imperfect scenarios, and how other humans respond to those decisions.

Armies of psychologists have spent entire careers attempting to understand human decision-making. It’s not easy to boil down to numbers and equations. Moreover, it’s conveyed through conversation, discussion and debate, something that technology has yet to harness or replicate.

Wargames have served as an indispensable tool in this exploration. They provide a way to exercise the decision-making process, explore why choices were made and determine what the implications might be. However, being human-centric isn’t always efficient. Wargames often require months of planning by experienced wargamers who deeply understand the defense issues at play. They also require human players with the expertise to emulate the various parties in a conflict. All this means wargames are often hosted on an annual cycle and can only explore a small number of the potential scenarios a national security leader might encounter.

After all, how many people could plausibly play the role of Vladimir Putin or Xi Jinping?

But with the advent of generative AI, we now have the ability to ask a computer to harness human language and, at a minimum, plausibly approximate human conversation and decision-making. That opens an opportunity to merge technology and wargaming in a way that hasn’t previously been possible — meaning we can bring wargaming to a wider audience over a broader set of possible scenarios.

Combine AI and physics-based modeling and simulation, which can traceably adjudicate how interactions between military platforms will play out (think whether or not an F-35 will be detected), and suddenly you can run wargames with a much smaller number of human players across a much larger number of scenarios. Because the artifacts of these games are captured digitally, you can then rapidly conduct assessments of exactly what happened and why it happened — which is incredibly labor-intensive in traditional wargaming.

While AI skeptics may rightfully point out that the future of AI has been overhyped for literally decades, we are no longer talking about the future of AI. It is a valuable addition to the wargaming toolkit right now — today. We know this because, with AI tools and a modeling and simulation backbone, we are building new scenarios in just days with a mix of AI and human players to explore numerous iterations, branches and variations of a conflict.

Are those AI players infallible? Far from it. But the ability to rapidly iterate the game — you can rewind and replay any move in a matter of seconds — allows you to explore a range of human and AI behaviors and begin to see the breadth of possible outcomes in any military scenario. And when you find that scenario that is critically important for human decision-makers to consider? That is when traditional wargaming really shines. Let national security leaders have that discussion and debate so that they can apply human judgment to the most consequential and important decisions.

The future of wargaming isn’t about traditionalists versus technologists. It’s about traditionalists and technologists working together, just like it was and is in professional baseball. We don’t need 20 years of debate to arrive at that conclusion.

Andrew Mara is the head of the National Security Analysis Department at the Johns Hopkins Applied Physics Laboratory (APL) where he leads the analytic team assessing the capabilities needed to solve the most pressing national security challenges; Kelly Diaz leads the Advanced Concepts and Capabilities program at APL, which aims to address complex national security challenges and inform strategic decision-making through innovative and data-driven approaches; Kevin Mather leads a team of analysts at APL in the development of advanced modeling and simulation analysis tools, including advanced framework for simulation, integration and modeling (AFSIM) and AI techniques to support complex national security analysis and decision-making.

Recognition of the looming decline has prompted allied countries to accelerate plans for icebreaker development, but before cutting steel, nations should ensure that next-generation designs address a wide range of future demands. Icebreakers are crucial for safety of life and constabulary missions in the poles, but their usefulness in naval warfare has historically been limited. If, however, new icebreakers were designed to deploy and service uncrewed systems, their mission set would greatly expand, bolstering a credible Arctic defense posture.

Heavy icebreakers are effective in areas with dangerous seasonal ice and hard multiyear ice. In the High North, they can break ice for convoys of resupply and merchant vessels through the Northwest Passage and Northern Sea Route, conduct rescue and salvage operations and provide presence for scientific and strategic purposes. Overlapping claims to the Arctic Ocean seabed around the North Pole by Russia, Canada and the Kingdom of Denmark may also encourage patrolling with icebreakers equipped for coast guard purposes.

To do so, icebreakers are fitted with powerful engines, sloping bows and reinforced hulls and propellers, enabling them to break and push ice. These features set them apart from other ships, but they also make icebreakers relatively slow, noisy and high in energy consumption. Additionally, construction of each vessel is expensive (up to $3 billion) and time-consuming, lasting 7-10 years.

Given their high procurement and operational costs, and crewing limitations, nations will face constraints on the number of icebreakers they can field. Additionally, some missions, such as anti-submarine warfare, require capabilities like sonar domes or arrays that are not well-suited to traditional icebreaker designs. But next-generation icebreakers could enable new approaches to these missions.

For example, to detect and track Russian submarines transiting from the Barents Sea to the Atlantic Ocean, NATO forces today use a mix of capabilities, such as satellites, seabed sensors, surface ships with towed arrays, maritime patrol aircraft and submarines. However, there are two primary limitations with the current approach. First, it does not scale well. It requires numerous crewed assets to cover Russian submarines, and if many Russian submarines simultaneously surge through the Greenland-Iceland-U.K. (GIUK) gap or other regions, there may not be enough forces to track and trail them. Second, Russian submarines can avoid many allied surface ships and aircraft by operating near and under the ice near Alaska, Canada and Greenland.

New icebreaker designs equipped with uncrewed assets can help solve these two problems. Fitted with helipads, cranes, moonpools and storage space, they can launch, recover and maintain airborne, surface and undersea uncrewed sensors and vehicles. Numerous uncrewed assets can help cover more area (in support of both civil and defense missions) and generate other effects, such as deploying sensors, laying mines, or, in the future, directly attacking submarines. To keep uncrewed assets operational, mothership icebreakers can refuel and maintain them — even in marginal ice zones, thus contesting Russian submarines under the ice. Additionally, by adopting large payload bays and interfaces, icebreakers can not only better support traditional missions such as navigation support, search and rescue and salvage today, but also be future-proofed to easily integrate with a range of uncrewed systems as they evolve to flexibly enable new warfighting concepts of operation.

Icebreakers cannot do it all. Allied forces will still need to field other assets across domains to efficiently operate, and if necessary, fight effectively in the polar regions. Accordingly, NATO states are wisely adding undersea sensors, uncrewed aircraft such as MQ-9B and crewed aircraft such as P-8A, ice-hardened frigates and polar satellites to their planned portfolios.

But icebreakers remain the lynchpin of Arctic presence. As allied countries consider new programs, they should look beyond safety of life and constabulary missions to maximize the value of icebreakers by adopting designs that also support broader defense demands. This week’s NATO summit is the perfect time to break the ice on adopting designs that leverage uncrewed systems from the start.

Liselotte Odgaard and Timothy A. Walton are senior fellows at Hudson Institute focused on Arctic, maritime, and defense issues.

NATO’s progress in the “final frontier” is notable, but insufficient given the degrading security environment, the breakneck speed at which the commercial space industry is evolving, and the high level of interest in space from allies. We believe that the next months will present a critical opportunity to advance NATO’s space mission and we offer five recommendations.

1. Update NATO’s space deterrence and defense policy playbook

NATO should update its 2019 Space Policy to better reflect the current threat environment and to complement the upcoming NATO Space Doctrine. The threat environment remains contested in Europe, on Earth and in space, amid concerns that Russia could put nuclear weapons in orbit or engage in other malicious activities. Additionally, China’s space activities, including the deployment of dual-use satellites with both military and civil applications, remain of concern.

NATO allies should also intensify efforts to develop internal response playbooks to a variety of Article 5 space scenarios and socialize them across the alliance via studies, wargames, and briefings. These playbooks would help map national and international authorities, capabilities, response options, red lines, and differences in allied priorities. The playbooks should explore several particularly thorny space contingencies: NATO’s transition from peacetime to conflict in case of a space-based crisis; NATO’s role in defending allied commercial space assets targeted by adversaries; NATO’s role in individual allied satellite and ground station protection; cross-domain deterrence and escalation; and potential targeting of adversary capabilities in third-party nations in response to an attack.

NATO may further benefit from extending these coordination efforts to select member and non-member states and socialize its thinking when preparing for specific contingencies. For example, NATO exercises could both learn from and reinforce Multinational Force Operation Olympic Defender. Additionally, NATO partners with significant space equities and synergies such as Japan could be invited to assist in exercises and discussions where relevant. These playbooks can help allies become more familiar with each other’s threat perception and develop “space IQ” on how to anticipate and coordinate actions, while maintaining the strategic ambiguity undergirding Article 5. While calls for establishing specific “red lines” against adversary action in space are understandable, we maintain that calculated ambiguity can support deterrence and retain allied sovereign political decision-making.

2. Implement NATO’s Commercial Space Strategy

NATO Assistant Secretary General for Operations Thomas Goffus noted that as a result of the explosion of commercial space capabilities, space is no longer “too expensive, too exquisite, and too classified.” NATO needs to understand how to employ these assets to enhance its military edge. Commercial space capabilities can enable states to satisfy mission requirements more efficiently, at lower cost, with potentially more opportunities for upgrades. NATO recognizes this dynamic. Soon, the alliance plans to publicly release its first Commercial Space Strategy.

While a valuable step in affirming the role of commercial space, the plan’s success will hinge on its strategy, and especially the model NATO develops for its engagement with industry. Past RAND research highlights several recommendations. First, NATO should develop robust partnerships and communication channels with space industry actors. Building on the positive example of SPACENET, a NATO platform for engaging space providers, NATO should continue bringing transparency to its communications and focus on including small- and medium-sized enterprises and non-traditional firms, which represent the backbone of allied defense industries.

Second, NATO should identify, authorize, and resource an institution to oversee engagement with commercial space providers. Currently, responsibilities are split among the NATO Space Operations Center, Space Center of Excellence, and others, which could lead to insufficient coordination.

Third, engagement with industry must be action-oriented to build trust and ownership. Initial conversations should seek ways to ease regulatory burdens that slow acquisition timelines. NATO and industry leaders should share information on the acceptable levels of risk in space operations, and how to build resilience for assured access. The use of commercial space capabilities is as much a cultural shift as a technical one. As Lt. Gen. Michael Guetlein recently said: “In the past, we didn’t believe we could count on our commercial partners and international partners during times of crisis. That has completely changed.” NATO’s Commercial Space Strategy implementation plan presents an opportunity to deepen this engagement and keep up with an industry that changes drastically every six months.

3. Fund the allied space enterprise

NATO allies have very tight budgets dedicated to space within their own military services despite the recent and expected increases in overall defense spending. Further, only around 0.3% of allied defense spending is dedicated specifically to NATO via common funding for items of shared interest such as operational costs, capabilities, and joint facilities. Common funding remains difficult to disburse, and many times requires a nation to “own” the start-up costs of even a jointly approved program. Rather, we suggest the model of project-based funding in support of multinational capability cooperation efforts as an alternative, such as the one established through the Alliance Persistent Surveillance from Space (APSS) program to integrate allied defenses.

EU efforts could also offer an additional source of funding. EU members have finalized the agreement for the SAFE €150 billion loan as part of the broader EU Commission package (ReArm Europe/Readiness 2030 Plan) to support the EU’s rearmament. Funding through the loan program could be used to expand and integrate European space capabilities by NATO states, including those who are members of the EU and those, like Norway, Canada, and the U.S., who are not but can participate in SAFE. Additionally, NATO may consider expanding cooperation with the European Space Agency, which is increasingly focusing on dual-use capabilities and redefining its defense role.

4. Develop interoperable space capabilities and effects

Ensuring credible deterrence means fielding appropriate capabilities. For the first time, this year’s NATO Defence Planning Process (NDPP) cycle will include space requirements for allied militaries. Notably, NATO is also shifting its space capability targets to an effects-based approach for each ally, focusing on quantifying the available “firepower” for allies. NATO should build on that progress by assisting allies to identify areas of joint interest for coordinated space capability development .

Second, allies should consider mainstreaming the multinational capability cooperation format, which allows for common procurement and acquisitions of capabilities in areas of immediate need. Increasingly, NATO is using a model of project-based coalitions to kick-start action in areas of immediate interest — NORTHLINK, STARLIFT, “Project Asgard”, Sky Fortress are positive examples, with APSS arguably the most successful one.

Seeded by Luxembourg with $17.7 million to launch the integration effort that would be available for all 32 allies, the program has now attracted more than $1 billion in allied contributions in the form of data, PED, and cash over the next five years. This model can be mainstreamed for particularly critical space enablers such as Satellite Comms, Intelligence, Surveillance, and Reconnaissance, and Command and Control, but also emergent capability areas such as space logistics, or In-Space Manufacturing and Assembly.

To maximize returns on investments, NATO allies should also identify where they can pursue joint, complementary capabilities over sovereign, redundant ones. For example, high-resolution imaging constellations easily fall under the former, whereas satellites used for nuclear command and control will certainly be owned and operated by individual states. Not all space capabilities can be categorized so easily, however, so there is a long list of assets for which discussion is warranted. Relatedly, while European allies are understandably concerned about their overreliance on U.S. defenses and security guarantees, an active effort to complement U.S. space constellations rather than replace them may remind U.S. policymakers of the benefits of allied cooperation in this domain.

5. Reaffirm NATO’s role as a space broker

NATO’s primary value proposition is to serve as a convening space for allies to discuss, debate, coordinate, and align defense activities. NATO leaders often promote global norms that benefit collective defense and strategic stability, such as responsible space behaviors. They may help national policymakers create technical and operational standards for developing and employing military space assets so that allies can invest in indigenous military systems while still maintaining interoperability. NATO can also be a catalyst to streamline intelligence sharing, conduct joint exercises, and facilitate capability development.

We argue that NATO should focus on what it generally does best and apply it to space by serving as a broker for the 32-member alliance across multiple dimensions. NATO should continue bringing its platform to bear in critical areas, such as setting space standards; building the “space IQ” among allied nations and their staff; enabling better coordination and readiness; and wargaming and exercising the space enterprise.

As the role of space grows in NATO’s architecture, a more forward-leaning role for the alliance may also be possible. However, this alternative route is significantly more complex. For example, proposals for “NATO-operated” space architectures are limited by extremely small funds available for pooled resourcing and significant political challenges for maintaining such multinational efforts.

NATO’s space efforts have evolved massively since NATO’s Space Policy was launched five years ago. We believe that by pursuing these five priorities, NATO can continue making progress in this critical domain.

Anca Agachi is the co-Lead of RAND’s Alliances Portfolio in the National Security Research Division, and a nonresident fellow with the Atlantic Council’s Transatlantic Security Initiative focusing on NATO issues. Shaan Shaikh is a defense analyst at RAND with expertise in air and missile defense. Retired U.S. Air Force Brig. Gen. Bruce McClintock is a senior policy researcher at RAND, where he leads its Space Enterprise Initiative.

Press coverage of TikTok might lead some to think that the platform’s capabilities and Chinese ownership pose a singular challenge to regulators and law enforcement officials in the United States, Europe and other democracies seized with concerns about protecting privacy, personal data and ultimately, their nations’ security.

The reality is that China’s premeditated path toward legally and illegally seeking and gaining access to Americans’ personal data began more than a decade ago when the PRC hacked and gained access to the Office of Personal Management’s (OPM) database related to security clearances, exfiltrating the sensitive personal data of more than 20 million U.S. citizens approved for access to classified government material. The PRC’s purpose in seizing the data of current and former U.S. officials does not require much imagination.

Since then, China’s methods have become more sophisticated and largely focused on commercial approaches leveraging the growing digital environment. Also, the vast distribution of software and hardware supply chains over the past 20 years, especially to China, have further facilitated the Chinese Communist Party’s (CCP) aims. Globalization ironically has been the grand enabler of President Xi’s vision to dominate the world economically and militarily on a foundation of stolen knowledge.

Technologies upon which the consumer market relies offer optimal access to sensitive personal data. Consider the payment systems in U.S. stores, where point-of-sale terminals made by PAX Technology, a Chinese company, have become widely used by banks and across the retail sector. In October 2021, the FBI raided PAX’s U.S. offices following reports of unexplained network activity and concerns about potential data vulnerabilities. A U.S. Treasury Department investigation later confirmed that PAX was sending encrypted data to unknown third-parties.

It’s not just financial information that is of interest to China. Access to Americans’ DNA through companies that are focused on health care and identifying ancestry appear even more ominous in the wake of the COVID-19 pandemic.

But setting individual data-rich companies aside, Chinese entities have also secured stakes in U.S. and allied cloud infrastructure platforms, raising questions about who ultimately controls or can gain access to sensitive enterprise and government data.

The challenge facing the U.S. and allied governments is how to protect systems, citizens, infrastructure and industries from China’s multi-front assaults. There are multiple approaches currently being used and new ones to be considered.

Imposing fines, such as the $600 million one imposed in May on TikTok by the Irish Data Protection Commission following claims that the app unlawfully transferred residents’ data to China without protection from government surveillance, is one method. Another is regulation on specific transactions. Interagency bodies such as the Committee on Foreign Investment in the United States (CFIUS) can effectively block a transaction involving Chinese entities acquiring U.S. companies with sensitive data.

The Executive Branch together with Congress are poised to continuously update the varied regulatory and legal regimes that touch on these matters. In addition to CFIUS and Export Control Reform Act (ECRA) authorities, there are Defense Department supply chain regulations focused on software and hardware supply chain security, integrity and resilience. These efforts are being expanded beyond the Pentagon to the Departments of Homeland Security, Energy, and Transportation, to cover critical non-defense-related data and infrastructure.

Market-based incentives — such as tax credits, loan guarantees, and preferential procurement approaches can effectively support domestic and allied alternatives to Chinese technologies. Fostering more opportunities and inducements for American innovation can help reduce dependence on foreign suppliers especially from risky countries.

But what about the Chinese companies that are already operating with impunity in the United States? Or corporate structures and state-directed investment funds that funnel money through seemingly non-Chinese-owned entities?

It may be timely to establish an interagency council concentrating on identifying high-risk threats, identifying gaps and coordinating mitigation strategies, thereby building on agencies’ work already underway. One option may be to consider means by which additional transparency requirements related to personnel and technology sources could be required – not only in relation to U.S. government contracts – but for non-Five Eyes foreign-influenced tech firms operating in the United States in critical infrastructure, data services, health care, and financial technology. To reinforce such an endeavor, America should deepen cooperation with its NATO allies and Indo-Pacific partners and collaborate in securing procurement pathways, as well as developing trusted platforms. Finally, focus is critical – assessing where China could gain a dangerous advantage and how best to leverage limited government resources.

The mission is vital. Not only do Americans deserve to know who controls the systems that collect and store their data, where that information might ultimately go, and for what purposes, they deserve to be protected from our adversaries’ malign actions. Safeguarding data is not enough. Across the board, from the hardware in our weapons systems to the software in our payment systems, the time to smartly reduce our exposure is now. Failing to do so won’t just cost us economically — it may one day cost us militarily. And by then, the price may be far higher than we are able to pay.

Mira Ricardel is a non-executive board director of Titomic and served as under secretary of commerce for the Bureau of Industry and Security from August 2017- April 2018 and then as deputy National Security Advisor April-November 2018. She was a vice president at the Boeing Company from 2006-2015.