As of today, ChatGPT-5 Pro is said to have an IQ of 148, as tested officially by Mensa Norway. It is now significantly smarter than most adult humans in the United States (who average 99.7). Grok 4 may be weeks away from becoming even smarter, but the progress at which AI reasoning inches ahead matters little when humans write code for these programs. However, AI has started to write its own code. In tandem, Mark Zuckerberg is building a super team dubbed the “superintelligence AI” lab and he offered a single person, Daniel Francis, $1.25 Billion for a four-year contract (or a $312 million per year salary). Further, Zuckerberg has gone on to poach the top AI talent from OpenAI, Anthropic, and Google, nearing 24 people in total out of a team of only 44.

Unfortunately, U.S. companies are also allowed to funnel money into Chinese AI companies, in part because it is a less expensive alternative than U.S. developed AI. China, as a near-peer adversary cannot be allowed to reach superintelligence first because whoever wins the AI race to superintelligence will have nearly unlimited computing ability and will be able to launch devastating cyber-attacks with ease.

If there are two teams approaching the finish line in a winner-take-all superintelligence race, then there is also a direct implication for long-term deterrence on global war. Imagine the following scenarios:

SCENARIO 1: The U.S. is ahead in the race to superintelligence, but China works diligently to steal code, launch cyber-attacks, and intimidate U.S. scientists. Eventually, China assassinates critical AI scientists, prompting the U.S. to threaten the use of nuclear weapons against China to stop its attacks. Yet, just before all-out war, China ceases its efforts, having become successful in its bid to cripple the U.S. AI industry so it can reach superintelligence first.

SCENARIO 2: The U.S. is ahead, but China is only barely behind. China uses its innovative AI models to wargame nearly unlimited sequences and calculates what it believes is the perfect attack to prevent the U.S. from reaching superintelligence first. In this scenario, the attacks never ramp up. Instead, it results in a massive, unprovoked first strike that incapacitates the U.S. This might be a nuclear strike or simply an EMP strike that decimates the U.S. power grid. Either way, China wins again.

SCENARIO 3: The U.S. and China hide their governments’ AI progress. Public companies continue progressing toward superintelligence, but one or both achieve it in a military or national laboratory behind closed doors. They ponder the best way to use it, leveraging it like the nuclear football in global diplomacy (i.e., setting the briefcase on the floor next to the President). They may have accessed superintelligence but lack confidence in the technology to use it for the near future.

SCENARIO 4: The U.S. and China hide their governments’ AI progress, and both achieve superintelligence behind closed doors. Then one day, one of them launches an attack on the other, prompting the other side to launch its own superintelligence response. The two AI agents battle across every sector of society, arm-wrestling for control. Seemingly trivial differences between one model and another let one win in one sector and the other win in another.

This article does not presume that the outcome of a superintelligence race is represented in one of these four scenarios. Rather, it argues that AI will inevitably complicate the landscape of deterrence as it may give confidence of victory in otherwise stable situations. This moment in history is nothing less than the moment when scientists Leo Szilard and Albert Einstein wrote President Roosevelt to warn of the potential use of fission in bombs.

The United States government must think carefully about the current state of AI in the world and what it will mean for deterrence strategy. We need to have a planned response if a superintelligence cyberattack is launched against the U.S. This includes physically isolating our command-and-control systems and planning for surprise attacks, itself planned by another country’s AI technology. Worse yet, military planners need to consider how to detect and respond to multiple grey zone micro-attacks that may be a component of a larger cascading attack.

We are amid our generation’s Manhattan Project moment. The 2023 Oppenheimer movie culminates in the detonation of the 1945 Trinity test. Perhaps if the United States plans well, in 80 years, we may all be able to enjoy a movie about Zuckerberg forming his superintelligence lab.

Rob Kittinger, PhD, is a Senior Fellow at the National Institute for Deterrence Studies. The views expressed are his own.

The AI Revolution’s Outsized Impact on Deterrence was originally published on Global Security Review.

Many assumed it was indeed the end of history. After all, between 1987–1993, Washington and Moscow signed four notable arms control deals: the Intermediate-Range Nuclear Forces Treaty (INF), the Strategic Arms Reduction Treaties I and II (START), the Treaty on the Prohibition of Nuclear Weapons in Outer Space, and the Conventional Forces Europe (CFE) treaty.

Russian strategic nuclear weapons were scheduled to drop from over 10,000 deployed to 3,500 by the year 2000. The INF treaty banned shorter range missiles altogether. And Warsaw Pact conventional forces in central Europe and Russia dropped precipitously.

President Reagan’s economic war against Moscow was successful. It ended the Soviet empire by pushing Moscow to the brink of insolvency. Russia could not financially maintain its formidable Cold War nuclear and conventional force levels.

From 1993–2001, the US did not enjoy the promised “end of history.” State sponsors of terror in Iran, Libya, Afghanistan, and Iraq took the fight to the US, albeit in a different mode than threatening to send massive tank armies through the Fulda Gap into Western Europe.

The US responded with a war that would last more than a decade and cost Americans an estimated $7 trillion. It was all for naught and accomplished very little.

Readiness and Modernization Shortfalls

While spending trillions on nation-building in Iraq and Afghanistan, the US Department of Defense (DoD) suffered from severe readiness and modernization shortfalls. The defense budget was roughly $305 billion in 1991 when the Soviet Union collapsed and almost exactly that in 2001 before 9/11. In the interim the budget dropped to as low as $250 billion and it was only after 1996 that the budget gradually increased to $300 billion.

When adjusted for inflation (1991–2011), the defense budget of $300 billion (1991), aside from “overseas contingency operations,” should have grown to $480 billion by 2011, assuming a 3 percent growth rate. That did not happen. The shortfall in defense spending reached $1.25 trillion during the two decades following the Soviet Union’s collapse.

The base defense budget in 2011 was roughly $500 billion, and at first glance equal to that expected. Out of a defense budget of $656 billion, $160 billion was allocated for the wars in Afghanistan and Iraq. In the two decades from 2001–2021, the Department of Defense spent $1.56 trillion on nation building—an average of $80 billion annually.

Over three decades after the Cold War’s end, the US did not invest in the modernization of the military. The three-decade peace dividend, which saw $2.8 trillion fewer defense dollars spent, was instead spent domestically and on nation building. As a result, the modernization and recapitalization of the armed forces, especially nuclear forces, were postponed.

By September 11, 2001, the US nuclear forces were already in the field for two decades (Ohio-class submarines), three decades (Minuteman III), and five decades (B-52). The nuclear budget, $77 billion at the end of the Cold War, dropped to less about $25 billion, with most of those funds simply maintaining legacy nuclear forces.

It was not until 2009–2010 that the Obama administration and Congress agreed on a plan for upgrading and replacing nuclear forces—three decades after President Ronald Reagan rolled out his nuclear modernization and sustainment plans in late 1981. New systems are projected to begin fielding in 2031 with completion by 2050.

The failure to prioritize the planning and implementation for replacing aging systems included nuclear command-and-control systems, warheads, and all three legs of the nuclear triad. The belief that the world was safer was a fool’s errand.

By shifting federal dollars from defense to social spending, the US also ensured the workforce needed to build nuclear weapons, space and missile defenses, and cyber systems are no longer there. Vendors associated with the building of Ohio-class submarines and the Minuteman III intercontinental ballistic missiles (ICBM) numbered in the hundreds once. Now, the nation is starting from scratch. The submarine industry lost 14,000 workers and now lacks the manpower to meet demand.

As for ICBMs, with the shutdown of the Peacekeeper production line, the US Air Force was left with a guidance and propulsion replacement program that over a period of more than a decade invested $8 billion in making sure the Minuteman III (1970) would stay in the force “through 2030.” Again, many hundreds of vendors no longer exist to make ICBM parts. Even worse is the current state of the available workforce. American universities grant more PhDs in the hard sciences to Chinese students than to American students. Across the board, the US has fewer workers in the hard sciences than needed, although industry is now reaching into the schools to bring students along a planned program of education that leads them to careers in the aerospace business.

The Challenge Ahead

The nation now finds itself in a precarious position at a time when China and Russia are at their most aggressive. The Columbia-class submarine, which will replace the Ohio-class submarine, was recently delayed two years, further increasing costs. And the herculean task of building 450 new ICBM silos armed with 400 missiles will prove costly. The US will maintain the current 400 ICBMs while simultaneously deploying 400 new missiles in new silos. The Sentinel ICBM, a technological marvel, is progressing toward production. It is a highly capable weapon that is planned for initial deployment in 2033.

Chairmen of the Senate Armed Services Committee, Roger Wicker (R–MS) recently concluded, “It will take several years of sustained investment and real growth beyond this down payment to keep pace with China’s military advances…. But to be clear: The cost of deterring war will always be dwarfed by the cost of fighting one.” This could not be more true. It is time the American people understand the challenge facing the nation and what it will take to overcome it.

Peter Huessy is a Senior Fellow at the National Institute for Deterrence Studies.

Invest, Don’t Spend, Peace Dividends was originally published on Global Security Review.

Golden Dome is focused on specific missile threats to the American homeland. AUKUS pillar two is designed to reduce the significant lead China has in dual-use emerging technologies. An “extended” Golden Dome approach that produces strong and resilient allies may provide greater strategic deterrence than an America alone approach.

As stronger allies contribute more to collective defense through burden sharing, this can reduce the financial and military burden on the United States. Capable allies can deter aggression and manage local conflicts, promoting regional stability without constant American intervention.

If allies are seen as weak and easily overrun, it may undermine the credibility of alliances and security guarantees, thereby emboldening adversaries. Strong allies often bring advanced technologies and capabilities that enhance joint operations through interoperability and innovation.

Defeating fractional orbital bombardment system (FOBS), intercontinental ballistic missiles (ICBM), submarine-launched ballistic missiles (SLBM), hypersonic glide weapons, and land-attack cruise missile threats is the focus of Golden Dome, which “proposes a multilayered defense network capable of intercepting threats during the boost, midcourse, and terminal stages of missile flight.” Some analysts argue that it is easiest to target these threats in the boost/ascent phase.

Targeting the boost phase can either occur from allied territory, targeting North Korean missiles from South Korea, or the homeland. The boost phase can also be targeted from space. However, targeting from space is not without its own unique set of challenges. Either way, it will not be solved by America without its allies.

Arguably, five-eyes countries (Australia, Canada, New Zealand, United Kingdom, and United States) are already entangled in American nuclear command, control, and communications (NC3). Furthermore, the US operates in coordination with other military forces as part of broader coalition operations. Hence, extending Golden Dome to allies is not only possible, but can further leverage the AUKUS pillar two effort. Moreover, a missile defense system consists of sensors, interceptors, and command-and-control systems that work together to detect, track, and intercept incoming missiles.

These necessary components exist in AUKUS pillar two working groups like the cyber capabilities, artificial intelligence (AI) and autonomy, quantum technologies, undersea capabilities, hypersonics and counter-hypersonics, electronic warfare capabilities, innovation and information sharing, and the deep-space advanced radar capability program (DARC).

AUKUS pillar two leadership should prioritize the development of technologies and supporting systems for an effective extended Golden Dome architecture. America is unlikely to solve the problem in isolation by building a “tightly integrated system of low Earth orbit (LEO) satellites, terrestrial radar stations, directed-energy platforms, and kinetic interceptors” that senses, decides, and neutralizes incoming missiles. The linkages and opportunities for the prioritization of advanced technology development for missile defense can be found in the following summaries.

Cyber capabilities encompass both offensive and defensive operations for missile defense. Offensive cyber tools are used to deter adversaries and disrupt their operations through tactics such as cyber-reconnaissance, communication isolation, targeted strikes, and network intrusions. Defensively, military forces can prioritize robust network protection, active threat disruption, and seamless coordination across units to safeguard critical systems. Cyber operations can enhance intelligence gathering, command and control, and information warfare to shape public perception and the broader information environment. As conflict evolves, training personnel in cyber tactics and integrating machine learning for threat detection and analysis can maintain strategic advantage.

AI and autonomy can transform missile defense and military operations by enhancing efficiency, precision, and decision-making. Autonomous weapon systems are used for reconnaissance, surveillance, and combat missions. AI-driven wargaming platforms simulate real-world combat scenarios to help strategists test tactics and improve readiness. In command and control, AI supports real-time data processing and analysis. AI optimizes logistics by improving resource allocation, supply-chain management, and transportation.

In intelligence and surveillance, AI analyzes vast datasets to detect patterns and identify threats. Additionally, AI monitors threats and predicts future events. Human-machine teaming allows AI systems to collaborate with human operators, combining strengths and minimizing errors.

Quantum technologies transform missile defenses through enhanced security, operational efficiency, and advanced training. In cybersecurity, quantum-resistant cryptography is being developed to protect against the threat quantum computers pose to traditional encryption. Quantum key distribution offers highly secure communication by transmitting encryption keys through quantum channels, making interception nearly impossible.

In military operations, quantum algorithms can optimize logistics, supply chains, and battlefield strategies by analyzing complex data in real time. Quantum computing could process data in real time, enabling missile defense systems to rapidly analyze incoming threats, allowing for quicker decision-making and more effective interception. Additionally, quantum computing could enable highly accurate simulations of complex systems like nuclear reactions and weapon designs.

Undersea capabilities encompass a wide range of offensive and defensive functions that contribute to missile defense. Offensively, submarines and other undersea platforms can strike surface vessels, submarines, and land-based targets. They are also instrumental in inserting special forces into hostile territory for reconnaissance or sabotage missions. Undersea vehicles play a key role in mine warfare, either by laying mines or clearing minefields. On the defensive side, these platforms are vital for anti-submarine warfare, enabling the detection and neutralization of enemy submarines. They also support surveillance and reconnaissance efforts, gathering intelligence on enemy naval movements. Undersea systems help protect vital infrastructure such as pipelines and communication cables and help ensure safe navigation.

Hypersonic weapons and counter-hypersonic systems are vital to missile defense operations. Hypersonic glide weapons and hypersonic cruise missiles are designed to strike targets with exceptional speed, maneuverability, and precision while evading traditional defenses. Counter-hypersonic capabilities include advanced sensors and tracking systems like radar and satellite imaging to detect and monitor hypersonic weapons. Hypersonic interceptors aim to neutralize threats mid-flight, while high-power lasers and microwave weapons can disrupt their guidance systems. Effective command-and-control systems are essential for coordinating these defenses, and soft-kill measures such as cyberattacks offer additional means to interfere with hypersonic weapons.

Electronic warfare is fundamental for gaining military advantages in cross-domain missile defense. Electronic attack includes jamming enemy communications, radar, and navigation systems. It also includes spoofing—sending false signals to enemy forces. Electromagnetic or directed-energy weapons disable or destroy enemy assets. Electronic protection ensures secure communication through encryption and satellite links and employs electronic countermeasures (ECM) to defend against attack. Electronic counter-countermeasures are used to overcome enemy ECM and maintain operational effectiveness. Electronic support can be focused on gathering intelligence through signals interception, using sensors for surveillance and target acquisition, and detecting threats in the electromagnetic spectrum.

            In short, under the auspices of AUKUS pillar two, Australia and the United Kingdom can contribute to Golden Dome in ways that many may not be thinking about. As longtime allies with a shared culture, history, and values, working together on Golden Dome just makes sense.

Natalie Treloar is a Senior Analyst at the National Institute for Deterrence Studies (NIDS), a Non-Resident Fellow at the Indo-Pacific Studies Center (IPSC), the Australian Company Director of Alpha-India Consultancy, and a cohost of the NIDS Deterrence Down Under Podcast.

Extending the Golden Dome: AUKUS Pillar 2 was originally published on Global Security Review.

This issue is not new. In his 1957 book Nuclear Weapons and Foreign Policy, Henry Kissinger made a controversial, yet analytically compelling, argument for the possible utility of tactical nuclear weapons in limited wars. He warned that massive retaliation was neither credible nor effective for deterring limited aggression and that a rigid dichotomy between conventional and strategic nuclear responses risked inviting coercion at the lower rungs of the escalation ladder. For Kissinger, introducing the possibility of limited nuclear use was not a call to war, but a recognition of strategic reality; the ability to escalate with restraint could deter adversaries from escalating first.

Fast forward to the 2030 scenarios modeled in Guardian Tiger I and II, and Kissinger’s insights remain disturbingly relevant. In the exercise, North Korea carried out a low-yield nuclear strike targeting South Korean naval vessels. American decision-makers, faced with the risk of horizontal escalation with China and the lack of consensus among allies, struggled to identify a proportional yet credible response. The idea of a retaliatory tactical nuclear strike was floated, but the simulated American leadership hesitated, reflecting both doctrinal ambiguity and an operational gap in American nuclear capabilities.

The risks of such hesitation are manifold. First, American restraint may be misinterpreted as indecision or weakness, particularly by allies like South Korea and Japan, who are directly exposed to North Korean and Chinese threats. Second, it creates an opening for adversaries to believe they can escalate to the nuclear level without inviting proportional retaliation. Third, it undermines the entire architecture of extended deterrence that underpins regional security.

Critics will rightly point out the perils of normalizing nuclear use. Introducing tactical nuclear weapons into a conflict zone invites moral hazards, increases the risk of miscalculation, and breaks long-standing nuclear taboos. It also challenges existing declaratory policies, such as the 2022 P5 Joint Statement affirming that “a nuclear war cannot be won and must never be fought.”

But these arguments, while valid in principle, must be weighed against the operational reality that a low-yield nuclear strike by an adversary may not be deterred by threats of massive retaliation. As the Atlantic Council report noted, North Korea’s nuclear doctrine increasingly incorporates elements of pre-delegated authority, tactical nuclear use, and efforts toward a more survivable second-strike posture. If the United States signals that it will not respond proportionally to a limited nuclear attack, North Korea may calculate that it can use nuclear weapons to coerce the South or constrain American action without triggering regime-ending consequences.

Moreover, the credibility problem is not confined to North Korea. China, observing Washington’s reluctance to respond in kind, may also be emboldened to engage in horizontal escalation, confident that the United States’s nuclear threshold is politically—and perhaps operationally—immobile. This perception could unravel the strategic coherence of integrated deterrence.

To address these challenges, US Forces Korea (USFK) and Indo-Pacific Command should adopt a more robust approach across multiple dimensions. First, the United States should consider forward-deploying platforms capable of delivering low-yield nuclear weapons. This could include the reintroduction of dual-capable aircraft or sea-based assets positioned in or near the Korean Peninsula. Such deployments must be both survivable and possess the ability to clearly signal an adversary of will, while being fully integrated into bilateral operational planning with the Republic of Korea (ROK).

Second, escalation options must be clarified through updates to American declaratory policy. This does not mean issuing public ultimatums or fixed thresholds but rather ensuring that adversaries understand the United States is willing to conduct proportional nuclear responses if deterrence fails. Strategic ambiguity must not become strategic paralysis.

Third, while the US and South Korea launched the Nuclear Consultative Group (NCG) in 2023 to enhance extended deterrence coordination, further institutionalization is needed. A structure modeled more closely on the North Atlantic Treaty Organization’s (NATO) Nuclear Planning Group would help deepen transparency, signal unity of purpose, and reduce the risk of fragmented responses during crises.

Fourth, both US and ROK forces must be equipped and trained to operate in the aftermath of a limited nuclear strike. This includes rehearsals and exercises focused on base survivability, radiological detection and decontamination, logistics continuity, and the resilience of command-and-control (C2) systems.

Fifth, strategic communication must be strengthened. Clear and consistent messaging to both adversaries and allies is critical. Deterrence depends not only on military capabilities, but also on the perceived credibility of those capabilities and the intentions behind them.

Ultimately, the goal of these measures is not to normalize the use of nuclear weapons, but to reinforce the threshold against their use by making deterrence more credible and responsive.

If that threshold is ever crossed and the United States fails to respond proportionately, the credibility of its extended deterrence architecture could unravel. The Guardian Tiger exercises highlight this grim possibility and should serve as a clarion call to action for policy and defense leaders alike.

As Kissinger warned in 1957, the danger of total war arises not so much from a deliberate decision to embark on it as from a series of actions which, though rational in themselves, cumulatively lead to disaster. The United States must ensure that its rational desire to avoid nuclear escalation does not lead to an irrational loss of deterrence. Tactical nuclear flexibility, responsibly exercised and credibly signaled, may be the painful but necessary insurance policy to uphold peace in East Asia.

Dr. Ju Hyung Kim, President of the Security Management Institute, a defense think tank affiliated with the South Korean National Assembly, is currently adapting his doctoral dissertation, “Japan’s Security Contribution to South Korea, 1950 to 2023,” into a book.

The Case for US Low-Yield Nuclear Options in Korea was originally published on Global Security Review.

Missile flight time between both India and Pakistan is a few minutes. Hence, commanders do not have time for verification and decide to launch a retaliatory strike on warning.

In Pakistan, a high-ranking Pakistani military officer contacts Pakistan’s prime minister to tell him India has initiated a nuclear strike, and an immediate response is required. The prime minister immediately orders a counterstrike. Pakistan proceeds to counterstrike protocols.

Missiles begin crossing borders between India and Pakistan as the world hurtles towards a nuclear catastrophe in South Asia. The world’s worst fears are realized. It is only later realized that the high-ranking Indian military officer’s message to an Indian commander was made by a nuclear command-and-control system embedded with generative artificial intelligence (AI).

The Role of Generative AI in Modern Warfare

The use of generative AI, which is best exemplified by ChatGPT, by an adversary, or third party, can impact deterrence stability between India and Pakistan. In a worst-case scenario, it could trigger a nuclear war, like the scene mentioned above. Should generative AI be embedded in nuclear command-and-control systems, it is theoretically possible that hyper-personalized messages, like the one described above, could be generated in a system that was not originally designed to allow such an outcome. The potential for a tailored message, for specific individuals, is possible.

Traditionally, content and information generation were based on human creativity. However, at present, generative artificial intelligence is being considered as a tool for digital transformation. Generative AI is a deep learning model which has the capability to generate synthetic media including text, video, and audio—dependent on training.

Perception Building and the Escalation of Crises

Generative AI can also play a distinctly different role in leading to nuclear crisis. In a deterrence framework, the development of a narrative and perception are crucial. Similarly, communication of deterrence threats is also achieved through perception-building of an adversary—concerning a state’s capabilities and credibility. In the specific case of India and Pakistan, India is creating misperceptions regarding Pakistan’s nuclear program. Recently, Pakistani Ambassador Munir Akram, at the United Nations Security Council’s high-level debate, stated that India is actively pursuing a state-led disinformation campaign to malign Pakistan.

He was referring to recent EU Disinfolab’s reports on how India is using fake nongovernmental organizations, think tanks, media outlets, and international organizations to malign Pakistan. Similarly, it is possible that India may use Generative AI to manipulate public opinion, thereby, creating misperceptions among Pakistan’s public. This could make the citizenry, as well as the government of both India and Pakistan, susceptible to separate facts from fiction—especially in a crisis situation. Hence, AI-driven narratives could push both countries to a crisis leading to warfighting.

Misinformation and Crisis Management Challenges in South Asia

Misinformation through generative AI can make crisis management and escalation dominance between India and Pakistan much more difficult. According to Heather Williams and Alexi Drew, Herman Kahn’s escalation ladder has transformed into a web across domains with multiple actors and alliances, especially in the age of social media. In South Asia, due to mated warheads and cannasterization of missiles by India, coupled with geographical contiguity with Pakistan, nuclear use may become ultimate escalation within minutes—without a ladder or a web. Social media could be a threat multiplier and can become lethal with the use of generative AI. Although, tweets are unlikely to independently start a crisis, tweets from government officials of India and Pakistan could be fabricated through utilizing hacking and generative AI to escalate an ongoing crisis.

Risks of Generative AI in Command-and-Control Systems

While Indian nuclear weapons may have an effective command and control where nuclear weapons have nuclear safety and security mechanisms, there was a “Brahmos missile crisis” in 2022. Inadvertent or accidental, the launch of a missile could trigger a reactionary strike by Pakistan as the missile is dual-capable. Furthermore, this fog of war could undermine strategic stability. As a Group Captain was blamed for the launch, many commanders like him, who may be capable of launching nuclear missiles, even on Indian nuclear submarines, could be manipulated via high-level orders created through generative AI.

False Flag Operations and Strategic Escalation

India has also opted for adventurism against Pakistan through its false flag operations and surgical strikes. Pakistan has adopted a conventional strategy of quid-pro-quo plus as a conventional response to such Indian limited strike. However, malicious false flag operations through disinformation and fake news by India to garner public support for electoral gains and domestic politics can result in crisis escalation in such a competitive strategic environment. It is due to mistrust within the adversarial relationship of India and Pakistan.

The Need for Confidence-Building Measures (CBM) in AI

There is a dire need for risk reduction and confidence-building measures (CBM) between India and Pakistan in the domain of AI. However, the appetite for CBMs and dialogue on the Indian side is non-existent. India and Pakistan could establish bilateral AI incident reporting and mitigation centers to counter the use of AI as a tool for inducing false alarms and other escalatory actions. Furthermore, there is a need for India and Pakistan to further opt for unilateral AI security and regulation measures, especially for inter-organizational and inter-state strategic communications. If leading states like the United States, China, and Russia opt for an AI arms control agreement, less powerful states, including India and Pakistan, may be persuaded to follow suit. This will help establish a universal AI arms control regime regarding the use of AI for military purposes.

Generative AI and Deterrence Stability: A Scenario for De-escalation

It is the night of May 28, 2028, at India’s military command center. A high-ranking Indian military officer sends a message to a subordinate commander; Pakistan has initiated a preemptive strike. Prepare for a retaliatory strike. This is not a drill.

Flight time between India and Pakistan is a few minutes. Hence, commanders do not have time for verification and need to decide quickly whether to launch a retaliatory strike. He calls the command center to verify the message received.

India’s prime minister is contacted immediately concerning the message received by one of his commanders. He picks up the hotline to Pakistan’s prime minister and asks, “Have you launched nuclear weapons?”

Pakistan’s prime minister responds, “Not at all. This is fake!”

India’s prime minister informs the military command center. The crisis is averted, and nuclear weapons are not used.

Confidence-building measures, which were present in the second scenario, have the opportunity to change the outcome of a conflict. Given the potential for artificial intelligence to be used in nefarious ways, it is time to take action to avert such a crisis.

Ms. Anum A. Khan is an Associate Director at the Center for International Strategic Studies (CISS) Islamabad. She is also currently a Research Fellow at Comprehensive Test Ban Treaty Organization and a Project Associate of The Third Nuclear Age Project led by Prof. Andrew Futter.

Generative Artificial Intelligence and Deterrence Stability between India and Pakistan was originally published on Global Security Review.