July/August 2026
By Maryann Cusimano Love

Pope Leo XIV calls for disarming artificial intelligence in his recent encyclical, Magnifica Humanitas. Although the U.S. government has largely abdicated a role in establishing AI safeguards, Leo issues a rallying cry. Civil society, citizens, AI builders, governments, and multilateral organizations should not leave pivotal decisions to the tech elites alone, he argues. Instead, he calls for everyone to play a role in building peace, protecting the vulnerable, and safeguarding the treasures of human relations from digital slavery.

The encyclical is an historic first: the first encyclical of Leo’s papacy, the first written by an American, the first by a mathematician, the first written in English, and the first focused extensively on AI. An encyclical is one of the highest forms of teaching in the Roman Catholic Church. It was addressed not only to the 1.4 billion Catholics who comprise the largest branch of Christianity, the world’s largest religion, but to everyone with the aim of advancing discussion and dialogue about urgent moral issues.

A New Long-Term Project

The Catholic Church is well-practiced in undertaking long-term building projects. The Milan Cathedral took nearly six centuries to complete. Leo just blessed the Sagrada Familia Basilica in Barcelona, after nearly a century and a half of construction, interrupted by the Spanish Civil War, World War II, and the COVID-19 pandemic. Hence, it is natural that Leo opens the encyclical with two contrasting building images that he returns to throughout the text: the negative example of building the Tower of Babel, a hubristic technological ascent, versus the positive example of the decentralized rebuilding of Jerusalem, a slower community construction project. He urges his readers to avoid “the Babel syndrome,” meaning an “idolatry of profit that sacrifices the weak,” and the dehumanization of reducing persons into data, performance, and targets to exploit. Instead, he advocates that people build together for the common good and assures that even if construction moves at a slower pace, a better result awaits.¹

In an unprecedented move, Leo spoke publicly about the encyclical when it was unveiled May 25, saying, “artificial intelligence needs to be disarmed. The word is strong, I know, but deliberately chosen.… The church has long been working for nuclear disarmament, aware that every great technical power can affect people’s lives and so must be accompanied by adequate moral discernment and public control. Nuclear disarmament remains a service to peace and the dignity of the human family.”²

He continued, “In a similar sense, artificial intelligence now demands to be ‘disarmed,’ freed from logics that turn it into an instrument of domination, exclusion and death. Like nuclear energy, it must be at the service of all and of the common good. Decisions about technology must never be separated from conscience and responsibility.… vigilance is necessary today. Peace, not merely the absence of war, is justice at work. But when technology weakens our critical sense, peace itself is at risk. Disarming, however, is not enough. We must build.… a future, not for a privileged few, but for the entire human family.”³

The encyclical is a broad and deep analysis of what it means to be human in a digital age, warning about advancing technology without a moral compass and the concentration of power in the hands of a select few tech elites. Leo underscores that “technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate and use it.”⁴ He notes the good that AI and emerging technologies can do in fields such as medicine, for example, but he also warns of the risks of technology being used to dominate, dehumanize, and exploit in pursuit of power and profit rather than the common good.

The encyclical peels back deeper layers beyond the usual policy questions over AI regulation and governance. Who are we and who do we want to be? What do we want to build to serve humanity? How can we build ethically, caring for one another and our common home? Leo contests the vision and building plans of some AI elites, such as Elon Musk, that view humans as merely the small bit of code and “biological boot loader” used to create superior machine intelligence.⁵ He covers the impact of AI on the environment, education, families, employment, human trafficking, truth and democracy, and of war and peace.

Why Listen to the Church?

Some tech powers in Silicon Valley welcomed Leo’s decision to open the conversation on the ethical concerns surrounding AI, which has become integral to war-fighting and national security decision-making. Christopher Olah, a co-founder of Anthropic, regularly consults Catholic thinkers on the ethics (or as he refers to it, the “soul”) of Claude and other Anthropic products; he participated in the release of the encyclical at the Vatican. Anthropic’s CEO, Dario Amadei, agrees with Leo that AI leaders such as himself should not be deciding the future of these technologies on their own.⁶

Others in Silicon Valley are pushing back, arguing that the pope should stay in his lane, and that the Catholic Church is not competent to discuss AI.⁷ Leo is a mathematician who leads the world’s largest nongovernmental educational network, including over 1,400 universities, 230,000 schools and educational institutions, academic institutes and think tanks on artificial intelligence and emerging technologies in almost all countries. They include The Catholic University of America’s Leonine Institute for Emerging Technologies and the University of Notre Dame’s DELTA network. Catholic professors, such as I, have degrees in AI and teach AI courses and programs. Catholics build and use AI programs, and Holy See diplomats negotiate on AI law and policy. Many Catholic priests, who are also scientists and engineers, serve as AI advisers to governments and international bodies, including the United Nations and European Union, and to the tech industry.

The Holy See led the Rome Call for AI Ethics, a pledge pioneering the development and application of common principles for AI ethics and regulation, including accountability, inclusiveness, impartiality, and privacy. Major tech companies such as Microsoft, IBM, Cisco, and Salesforce are signatories. The Pontifical Academy of Sciences traces its lineage from one of the oldest scientific forums in history, the Accademia dei Lincei (1603-1651), which counted Galileo Galilei among its members. Many Nobel laureates are members, and AI has been a dialogue theme between religious and scientific leaders for many years now.

Nobel laureates and Catholic experts met July 14-16, 2025, in Rome for a discussion of AI and nuclear weapons. For more than a decade, long before the advent of ChatGPT and today’s large language models, the Catholic Church has been convening the Minerva Dialogues between AI industry leaders and religious leaders, urging that AI respect human dignity and serve the common good. Two Church offices, called dicasteries, released Antiqua et Nova: Note on the Relationship Between Artificial Intelligence and Human Intelligence last year. Leo’s Magnifica Humanitas encyclical is not a one-off; it is the fruit of this long engagement and discernment on these issues.

Realist thinkers argue that religious actors and factors have no role in international relations or should have no role,⁸ but this assertion rather proves the point that the pope is making about the need to overcome a culture of exclusion. It is not merely a matter of creating more moral machines. “A more moral AI is not enough if that morality is determined by a few,” Leo wrote.⁹

Unlike the frontier AI companies, the Catholic Church is not selling AI, so it can speak frankly. The tech titans have a conflict of interest in promoting their lucrative interests while claiming there are no serious downsides to their products, tech companies can be trusted to self-regulate, yet resisting attempts to strengthen requirements for transparency, accountability, safety regulations, and guardrails. Palantir co-founder Peter Thiel—a self-identified libertarian, conservative Christian, MAGA contrarian—seems to be trolling Pope Leo. In a series of lectures held in Rome near the Vatican, Thiel claimed that those who propose AI regulations are the Antichrist.

Disarming AI

Leo goes beyond calls for AI regulation and safety. He urges deeper relationships for deeper disarmament and says that any technological advancements detached from ethics and human responsibilities exacerbate, accelerate, and dehumanize conflict, cyberattacks, and hybrid war, rather than building just peace. “Disarming AI means freeing it from the mentality of ‘armed’ competition, which today is not limited simply to the military context, but is also an economic and cognitive phenomenon. This entails a race for ever more powerful algorithms and larger datasets, driven by the desire to secure geopolitical or commercial dominance,” he writes.¹⁰

Leo continues: “To disarm means discrediting the assumption that technical power automatically confers the right to govern. To disarm does not mean rejecting technology but preventing it from dominating humanity. It means freeing technology from monopolistic control and opening it to discussion and debate, therefore making it human-friendly and restoring it to the plurality of human cultures and ways of life.… For this reason, merely regulating it is insufficient; it must be disarmed, welcoming and accessible.”¹¹

In this context, the pope says that disarming AI means confronting realpolitik, the culture of power that normalizes war under a false realist school of international relations theory that incorrectly claims conflict is inevitable, so therefore arm up. By erasing truth and the historical memory of the harms and horrors of war, algorithms and communications networks reward conflict and stoke disinformation and fear while sanitizing and gamifying violence.

In the encyclical, Leo expands upon previous Catholic calls against lethal autonomous weapons systems and in favor of nuclear disarmament and arms control. He notes that “No algorithm can make war morally acceptable,”¹² and sets out three non-negotiable requirements for the use of AI in military systems. First, the buck must stop with a person; accountability and blame cannot be “collapsed into ‘the machine,’” but must be traceable to human decision-making processes. Second, the use of force must “remain under effective, self-aware and responsible human control.” And third, at the communal level, the world must develop a shared international framework to curb the AI and technological arms race.¹³

Leo further makes the point that words are powerful and must be chosen wisely to favor more peaceful, less aggressive language over confrontational language that divides humanity into in-groups and out-groups.” He urges people to disarm hearts, words and relationships, in order to build a just peace and move away from the normalization of war, the arms race, the false logic of deterrence, and the just war tradition.

Just Peace Over Just War

In repudiating the normalization of war, Leo criticizes just war theory as outdated. Some media outlets jumped on this quote and reported that the pope had overturned just war tradition. Not exactly.

The just peace tradition is the majority position within the church and holds that the objective at all times is to prevent conflict, end conflict, and build peace. The just war tradition is the minority position.¹⁴ Popes have emphasized the just peace tradition for decades, while de-emphasizing the just war tradition. Leo clears up several common errors. First, he says that Jesus worked to build a just peace without violence. The just war tradition was never the foundational Catholic teaching on war and peace. It emerged centuries later to limit violence, not to justify or expand war, as it is commonly misunderstood today.

Second, Leo notes that St. Augustine has been stripped of his original emphasis on building a just peace and turned into a caricature as a proponent of war. An example is when Vance tried to invoke Augustine to justify the U.S. attacks on Iran—Leo is not having this. Throughout the encyclical, he writes that the Catholic tradition seeks to build a just peace, while lamenting the false manipulation of the just war tradition to justify any kind of violence. Unjust wars are being fought for profit, and arms makers and the military-industrial complex fuel conflicts to reap these profits.

Leo reiterates that “preventive” attacks are not just, and do not meet the strict conditions of self-defense, nor do unjust wars fought to divert public attention from domestic problems. As the former head of the Augustinian order (Catholic religious orders are often named after the saints that founded or inspired them) and a scholar of St. Augustine, Leo reminds his readers that the saint put the drive for a just peace first and foremost among his priorities. “In an increasingly interdependent world, peace is not just one issue among others, but a prerequisite for the universal common good and a test of the moral maturity of peoples, especially of those who bear responsibility for governing,” he writes.¹⁵

Third, Leo notes that humanity has far more effective tools to resolve conflict, such as dialogue, diplomacy, and forgiveness, than its leaders have sought to exercise. He recommends applying all the just peace principles and practices—expanding participation in restoring communities, building right relationships, truth-telling, and reconciliation—to build a more sustainable peace with one another and with our common home. These tools are practical and have a successful record, which Leo documents: restoring and rebuilding Europe after World War II and building multilateral and international laws and institutions, such as the UN, the Universal Declaration of Human Rights, the Refugee Convention, the Treaty on the Prohibition of Nuclear Weapons, the U.S. civil rights movement, the anti-apartheid movement in South Africa, and Wangari Maathai’s Green Belt movement. With more effective tools available to build just peace, the just war doctrine is outdated.

Finally, given the immense destructive potential of modern war, Leo writes that the “evils” of war outweigh the harms that force seeks to eliminate. He does not mention the U.S.-Israeli war on Iran specifically but notes that, “It is much easier to start a war than to stop it.”¹⁶

All popes since World War II have called for a greater emphasis on building a just peace, noting that almost all conflicts do not meet the criteria for a just war. Catholic peace groups such as Pax Christi USA and Dorothy Day’s Catholic Worker call for going farther by totally eliminating the just war tradition, in the same way that the Catholic Church changed its position from supporting to opposing the death penalty. It is more likely that Leo the Augustinian will restore St. Augustine’s emphasis on just peace, rather than jettison the just war tradition entirely. He mentions “peace” 67 times in Magnifica Humanitas, and “just war” only once. He retains the narrow strict exception for self-defense, as in the case of Ukraine’s self-defense against Russian attacks, and calls for increased protections of civilians, consistent with the just war tradition’s in bello criteria.

Leo disputes that peace is utopian.¹⁷ Empirically, it is not. Most of the world is at peace, most of the time. We work to build peace in our families, workplaces, and communities. Leo is calling humanity to the construction site and telling us to bake bricks and bring them along. Building peace is a moral and religious obligation and a practical action plan.

Criticisms of Nuclear Deterrence

Some proponents of nuclear deterrence hoped Leo would overturn his predecessor’s encyclical, Fratelli Tutti, which holds that the possession of nuclear weapons and adherence to nuclear deterrence are immoral. They will be disappointed. There is no daylight between Leo’s critiques of nuclear deterrence and those of his predecessor, Pope Francis. The new encyclical calls out “the widespread yet erroneous belief that nuclear deterrence is an indispensable prerequisite for security.”¹⁸ From a just peace perspective, deterrence displays “a relational poverty that always has disastrous consequences for civilian populations,” Leo writes.¹⁹

None of the nuclear armed states are engaged in disarmament negotiations, and the Trump administration argues that this is not the time for such negotiations, that diplomacy must wait until conditions are ripe. Instead the administration prefers bilateral consultations on risk reduction and nuclear testing. Leo disputes this narrative and any assertion that there are parties with whom it is illegitimate to engage. In his view, the time is always right for disarmament negotiations, diplomacy, and dialogue. Leo invokes Fratelli Tutti, which also notes that it is immoral to possess nuclear weapons.²⁰ Dialogue is not transactional but transformative, Leo writes. Diplomacy is a vocation, requiring abundant perseverance, humility, patience, and openness to the other.

For those who hope the Catholic Church will bless the bombs that undergird nuclear deterrence as the Russian Orthodox Church does, that is not going to happen. Not all remarks made by the pope are authoritative, binding, high-level teaching of the Catholic Church. If Leo roots for the White Sox and orders a hot dog at the game, Catholics are not bound to do the same. At the top level of Catholic teaching are apostolic constitutions (such as Vatican II), followed by encyclicals. At the end of the line are pastoral letters by just one bishops’ conference and remarks. The calls for abolishing nuclear weapons, nuclear disarmament, and the critiques of deterrence are in the constitutions and papal encyclicals of higher levels of Catholic teaching. The 1983 pastoral letter of the U.S. Catholic bishops, which in one section offered limited strictly conditional acceptance of nuclear deterrence as long as it was on the path to deeper disarmament,²¹ was lower-level Catholic teaching, now outdated.

Leo proposes several ways forward: disarming words by speaking the truth; building peace in justice; adopting the perspective of victims; and rejecting realpolitik and seeking practical paths of peace through deeds, diplomacy, dialogue, and multilateral cooperation. Even in the darkest nights, peacemakers build and persevere with astonishing creativity in doing good, protecting the vulnerable, and opening pathways to reconciliation. We are invited to the construction site to build a more fully human future in the digital age, and not to give up just because big tech owns the brick company.

ENDNOTES

1. Pope Leo XIV, Magnifica Humanitas, May 25, 2026, paragraph 10. All following quotes from Magnifica Humanitas include just the paragraph number in brackets.

2. Pope Leo, Remarks at the public event releasing Magnifica Humanitas, May 25, 2026, Vatican City.

3. Ibid.

4. Magnifica Humanitas, paragraph 9.

5. Elon Musk, “Hope we’re not just the biological boot loader for digital superintelligence. Unfortunately, that is increasingly probable,” Twitter, August 3, 2014.

6. Elias Wachtel, “Why Silicon Valley is Turning to the Catholic Church,” The Atlantic, April 25, 2026.

7. Cade Metz, “At the Epicenter of A.I., Pope Leo’s Warnings Are Dismissed,” The New York Times, May 26, 2026.

8. Maryann Cusimano Love, Global Issues Beyond Sovereignty, Chapter Five, “Religious Actors and Factors in International Politics” (Rowman & Littlefield: Lanham, MD, 2020).

9. Magnifica Humanitas, paragraph 107.

10. Magnifica Humanitas, paragraph 110.

11. Magnifica Humanitas, paragraph 110.

12. Magnifica Humanitas, paragraph 198.

13. Magnifica Humanitas, paragraph 200.

14. Maryann Cusimano Love, “Just Peace and Just War,” Expositions, Vol. 12, No. 1 (2018), pp. 60-71.

15. Magnifica Humanitas, paragraph 182.

16. Magnifica Humanitas, paragraph 195.

17. Magnifica Humanitas, paragraph 205.

18. Magnifica Humanitas, paragraph 194.

19. Magnifica Humanitas, paragraph 192.

20. Maryann Cusimano Love in Chris Seiple and Dennis R. Hoover (eds.), Routledge Handbook of Religious Literacy, Pluralism, and Global Engagement, Chapter Four, “Fratelli Tutti: Lessons Learned from Interreligious Action and the Catholic Church” (Taylor & Francis, 2021).

21. U.S. Conference of Catholic Bishops, “The Challenge of Peace: God’s Promise and Our Response” pastoral letter, May 3, 1983.

July/August 2026
By Tianjiao Jiang

The 2026 nuclear Nonproliferation Treaty (NPT) Review Conference once again unfortunately failed to reach a consensus final document. The Iranian nuclear issue is certainly a major reason, but the competing views between China and the United States on nuclear arms control have also been fully reflected in the review process outcome. China has firmly refuted U.S. criticism about the modernization and opacity of the People’s Liberation Army’s nuclear arsenal and rejected U.S. multilateral nuclear disarmament and dialogue initiatives.¹ In Beijing’s view, the U.S. multilateral negotiation initiative is an attempt to shirk its responsibility as a nuclear superpower to prioritize nuclear disarmament, while criticism of China’s nuclear modernization is an excuse for its own nuclear expansion.²

As the review conference convened in New York, the leaders of China and the United States met in Beijing and established a new bilateral relationship centered on constructive strategic stability.³ Theoretically, this approach should include actively engaging in dialogue and cooperation on nuclear arms control and nonproliferation issues, avoiding arms races, and strengthening crisis management. However, Beijing and Washington have not provided a clear path. Therefore, this article attempts to explore from a Chinese perspective the long-term structural challenges faced by China and the United States on nuclear strategic stability, and to provide suggestions for advancing further bilateral engagement and dialogue.

Recent China-U.S. Nuclear Discussions

China recently released its first arms control white paper in 20 years.⁴ In addition to reiterating long-standing policies such as no first use of nuclear weapons, nonparticipation in arms races, and ensuring a defensive national defense policy, it included a chapter on arms control in emerging technologies—space, cyber, and artificial intelligence—thus reflecting China’s willingness to play a more active role in multilateral arms control. The white paper also explicitly states, however, that only when the countries with the largest nuclear arsenals take the lead in nuclear disarmament and make further substantial cuts in their nuclear arsenals will conditions be ripe for all states to participate in multilateral nuclear disarmament.

In recent years, the U.S. strategic community has expressed serious concerns about China’s rapidly growing and opaque nuclear arsenal. The Pentagon’s 2022 China military report estimated that China’s nuclear weapons stockpile will reach 1,500 by 2035,⁵ although a subsequent report put the estimate at 1,000 nuclear weapons by 2030.⁶ Regardless, the trajectory has generated serious consideration about how to simultaneously confront both China and Russia as strategic nuclear competitors in the so-called tripolar nuclear age.⁷

Meanwhile, the United States has been criticizing China’s refusal to participate in trilateral arms control negotiations. Some experts argue that because China lacks the will for arms control, the United States should not continue engagement but instead shift to a more competitive policy to force China to reconsider its position.⁸ This could include encouraging so-called friendly nuclear proliferation—deploying or even allowing its Asia-Pacific allies to possess nuclear weapons⁹—as well as further strengthening U.S. nuclear forces, developing a nuclear-armed, sea-launched cruise missile, strengthening a conventional triad and forward deployment of military units to deter China.¹⁰

Such practices would not resolve the structural contradictions in U.S.-Chinese strategic stability. On the contrary, they may provoke China to take even tougher measures, leading both sides into a more serious security dilemma and conflict escalation. China’s reluctance to respond to U.S. arms control proposals stems primarily from fundamental differences in nuclear strategy between the two adversaries. China’s arms control white paper clearly states that China has long adhered to the principle of no first use. Although this principle is often questioned, it reflects Beijing’s consistent belief that nuclear weapons are political weapons, whose role is limited to deterrence or counter-nuclear blackmail and cannot be used in war-fighting. China’s nuclear principle emphasizes a small, lean, and efficient force. As U.S. strategic scholars such as Robert Jervis have argued, as long as a country possesses reliably survivable and credible nuclear retaliation capabilities, nuclear deterrence is automatically effective, regardless of the size of the nuclear arsenal.¹¹ Therefore, China vows not to participate in any arms race.

The U.S. nuclear strategy is completely different. Due to the Cold War rivalry between the United States and the Soviet Union, the U.S. nuclear strategy focused on war-fighting. Since the 1960s, the United States not only deployed thousands of tactical nuclear weapons but also developed numerous theories on escalation dominance and formulated a series of nuclear strike plans. In the U.S. view, nuclear weapons could be used on the battlefield, and Washington had to win against Moscow by cleverly controlling the use of nuclear weapons.¹²

Although the United States and the Soviet Union quickly fell into a state of mutually assured destruction, the U.S. government and strategic community sought ways to escape this situation of mutual hostage-taking. Thus, counterforce and damage limitation strategies emerged, aiming to destroy the opponent’s nuclear forces through a first strike on the one hand, and to intercept the opponent’s nuclear retaliation and reduce casualties through missile defense, on the other.¹³ During the Cold War there were constraints imposed by the 1972 Anti-Ballistic Missile (ABM) Treaty, but U.S. President Ronald Reagan’s Strategic Defense Initiative was not technologically feasible, and successor initiatives have only been partially effective.

China’s Nonresponse to Nuclear Negotiations

For China, the prospect of the U.S. nuclear war-fighting and damage limitation strategy is deeply unsettling. After the end of the Cold War, the United States quickly withdrew from the ABM Treaty and began accelerating the expansion of its global missile defense system, culminating in the recent Trump administration proposal for the “Golden Dome” plan. According to experts who participated in U.S.-Chinese Track II arms control dialogues over the past few decades, China has complained consistently to the United States about its missile defense expansion, but Washington has been unwilling to impose any limitations.¹⁴ Because China until recently only possessed 200 to 300 nuclear weapons, a U.S. first strike, combined with missile defense interception, could potentially prevent China’s nuclear retaliation from hitting its targets. This concern is corroborated in papers by Keir A. Lieber and Daryl G. Press, who noted that around the year 2000, the United States did not believe China possessed a reliable nuclear retaliatory capability and therefore refused to accept the concept of mutual vulnerability between the two countries.¹⁵

By 2010, with the development of China’s military capabilities, Beijing and Washington began discussing strategic stability, but these discussions did not last. One significant obstacle was the U.S. alliance system and extended deterrence strategy, which had been in place since the Cold War. Beijing, having learned from the Sino-Soviet split that alliances could be costly, maintained a nonaligned stance and opposed providing a nuclear umbrella to other countries. During the Clinton administration, China and the United States reached an agreement on mutual non-targeting of nuclear weapons, which Beijing saw as an indication of an improving strategic relationship.¹⁶ However, China was soon put back on the U.S. nuclear strike list.

Beijing believes that as long as Washington is unwilling to discuss the no-first-use principle, it means that the United States intends to launch a first strike against China and offset its retaliation capability by damage limitation. During the Obama and Biden administrations, U.S. officials discussed the possibility of adopting a “sole-purpose” nuclear doctrine, which is similar to a no-first-use policy. This was not possible due to opposition from East Asian allies such as Japan, which believed such a policy declaration would weaken the U.S. extended deterrence. Japan argued that a U.S.-Chinese strategic stability agreement would also reduce the credibility of Washington’s security guarantees to Tokyo.¹⁷

From China’s perspective, structural contradictions in the U.S.-Chinese strategic stability relationship have long existed. Beijing has raised many topics for discussion, but Washington has been unable to respond for various reasons. Now, critics in the United States question why, despite these long-standing issues, China suddenly significantly expanded its nuclear arsenal and deployed various advanced weapons, such as hypersonic weapons, at some point in the past decade.

The answer is obvious. During his first term, President Donald Trump explicitly advocated a return to great-power competition, designating China as the strategic rival and pacing threat, demanding an increased role for nuclear weapons, deploying the W76-2 low-yield nuclear warhead—which is understood to be aimed at China and blurs the line between nuclear and conventional weapons—and insisting on intercepting missiles fired at the United States from anywhere at any time. The last point marks a major expansion in U.S. missile defense policy because it includes intercepting China’s nuclear retaliatory forces and thus undermining the credibility of China’s nuclear deterrence. Coupled in Trump’s second term with a trade war, a tech war, and the Taiwan issue, mutual trust between Beijing and Washington has sharply declined.

To adapt to the return of great-power competition, the U.S. strategic community advocates winning the new cold war through technological and arms races. In this view, the United States never accepted the notion of mutually assured destruction with the Soviet Union in the past and should not accept mutual vulnerability with China now. It holds that the United States needs to ensure absolute military superiority through the development of emerging technologies and offset its adversaries’ retaliatory capabilities.¹⁸

This perspective is also reflected in relevant U.S. military operational concepts and actions. They include the U.S. Cyber ​​Command’s “persistent engagement” and “hunt forward” strategies, which could potentially penetrate China’s nuclear command-and-control systems and disrupt Chinese missile launches. In Beijing’s view, the Starlink and Golden Dome programs could become space weapons. The Starlink program plans to launch more than 40,000 satellites, while China currently has fewer than 1,000 satellites in orbit. Starlink could threaten China’s space assets with a collision—China’s space station has been approached by Starlink satellites several times without knowing their intention—and support preemptive actions by the U.S. military. Through the use of artificial intelligence, the U.S. military can not only significantly improve the effectiveness of missile defenses such as Golden Dome, but also more effectively identify, locate, and track China’s mobile missile launch vehicles by more efficiently analyzing satellite imagery, electromagnetic signals, and cyberattack intelligence. All of this poses a serious challenge to the credibility of China’s nuclear retaliation.

To restore strategic stability and enhance the credibility of its nuclear deterrent, China is making dynamic adjustments to its nuclear arsenal. The United States criticizes this approach, arguing that if the goal is to improve survivability, China could simply increase its strategic nuclear submarines instead of increasing missile silos. This approach would be difficult because strategic nuclear submarines require a high degree of stealth technology, which China cannot rapidly increase in the short term. Expanding the number of missile silos, by contrast, would buy China time and consume U.S. strike capabilities and attention.

Beijing’s reluctance to engage in dialogue with Washington on this issue is largely due to differences in strategic culture. From the U.S. perspective, the lack of mutual trust makes strategic dialogue between the two countries even more necessary, and they should discuss various nuclear issues separately, following the compartmentalized approach employed by the United States and the Soviet Union in the Cold War. However, China is not the Soviet Union and insists on viewing this issue through a comprehensive security framework. Without a foundation of mutual trust, and divided over other major issues such as trade and Taiwan, effective dialogue between the two countries is very difficult.

Can the U.S. Engage China on Arms Control?

Currently, the U.S. strategic community is relentlessly criticizing China and hyping a nuclear tripolar rivalry, arguing that China and Russia will force the United States to rely on nuclear capabilities in two theaters, either at the same time or sequentially, and using this as a pretext to strengthen nuclear weapons modernization and conventional triad forces.

On the one hand, this viewpoint is highly misleading. Although the United States appears to be cornered, China and Russia have not actually become military allies. Why does the United States not consider a divide-and-conquer strategy against China and Russia, instead deliberately creating a more difficult situation that would involve simultaneously contending with both?

On the other hand, this kind of narrative not only fails to resolve the long-standing structural contradictions in U.S.-Chinese strategic stability but also intensifies the arms race and security dilemma. For the United States, its massive nuclear arsenal overhaul program is likely to be costly and may not even achieve the desired results. Historically, such projects have often been behind schedule and over budget. For example, a key component of the 2018 Nuclear Posture Review was the replacement of land-based intercontinental ballistic missiles, originally planned for completion in 10 years, but now likely to be delayed until 2050 at best, with budget costs projected to nearly double.

The United States also needs to simultaneously build new plutonium pits, new warheads, new bombers, new missiles, and strategic nuclear submarines, but has stopped manufacturing nuclear weapons for over 40 years. Restarting this manufacturing operation will impose enormous time, money, manpower, and environmental costs.

Why is the United States wasting resources and employing outdated Cold War methods to address problems 80 years later, with results that are likely to be unsatisfactory? Why not consider resolving the long-standing structural contradictions between China and the United States through cleverly designed arms control mechanisms and gradual adjustments to nuclear strategy? Although the United States criticizes China for being unwilling to engage in arms control dialogue, the United States never took Chinese concerns seriously.

In many Track II dialogues, U.S. experts have proposed numerous ways to reduce nuclear risks, including increased transparency, missile launch notification, and nuclear fail-safe, but China views acknowledging the mutual vulnerability between Beijing and Washington as the primary concern. Without addressing this concern, other risk reduction measures are merely stopgap and could even have serious negative consequences. That is because China has always worried about a preemptive U.S. strike and believes that without acknowledging mutual vulnerability, increased transparency and information sharing could easily be exploited to issue nuclear threats or carry out a decapitation strike. The United States should demonstrate to China the tangible benefits of participating in arms control dialogues, such as exploring mutual vulnerabilities and the sole-use doctrine to prevent escalation of nuclear conflict. Trump’s explicit expression of aversion to nuclear war suggests that he would also be interested in these topics.¹⁹

The establishment of a constructive strategic stability relationship would provide an opportunity for China and the United States to engage in a nuclear strategic stability dialogue. The May 17 summit in Beijing did not explicitly link constructive strategic stability with nuclear strategic stability. Considering that there will be multiple other opportunities for the top leaders of China and the United States to meet this year, the strategic advisers on both sides should work together on the following issues.

First, they should convey clear messages to each other’s top leaders that a strategic stability dialogue should be put on the table as one of the priorities. Due to the interference of numerous other issues in bilateral relations, strategic stability has long failed to receive the attention it deserves. Given the political polarization in the United States, China is also keen to know how the United States views the issue of strategic stability with China. What might the future U.S. strategic direction look like? The Trump administration has also clearly stated its desire to avoid a major military conflict with China through dialogue.

Next, the strategic advisers should suggest that the two leaders reaffirm in a joint statement that a nuclear war cannot be won and must never be fought, thereby fostering mutual trust and setting an example for strengthening the global nuclear taboo and reducing nuclear risks. This commitment should already be implicit in the constructive strategic stability relationship, but both sides are still building and have not truly established a strategic stability relationship. Such a statement can be seen as an indirect way of saying that the mutual vulnerability exists, thus avoiding pressure from U.S. domestic politics and allies. China and the United States should also jointly conduct strategic dialogues with stakeholders in the Asia-Pacific region to explore how to achieve regional security through conventional measures and legal frameworks instead of enhancing the role of nuclear weapons.

Finally, both sides should engage in dialogue on issues such as nuclear doctrine, missile launch notifications, transparency and confidence-building measures, nuclear fail-safe, and crisis communication. China and the United States recently resumed a bilateral dialogue on AI, but mainly discussed issues related to civilian technology regulation and export control. Placing a special emphasis on the entanglement of emerging technologies such as AI, cyber and space with nuclear weapons, both sides should quickly establish terminology handbooks and risk management channels through joint research and dialogue mechanisms to prevent accidental nuclear escalation due to technical flaws or misunderstandings.

In September this year, China and the United States will hold another bilateral summit. Experts from both sides should work together toward these goals and strive to solve the structural problems that have long plagued the strategic stability of the two countries. If this opportunity is missed, the logic of strategic competition may overwhelm the notion of constructive strategic stability, leading to a further decline in strategic mutual trust between the two sides and a deeper security dilemma. It is crucial for international peace and stability that China and the United States do not engage in a nuclear war and avoid the escalation of nuclear conflicts. This is a red line that cannot be crossed in the competition.

ENDNOTES

1. Sun Xiaobo, “Statement at the General Debate of the Eleventh Review Conference of the Treaty on the Non-Proliferation of Nuclear Weapons” (in Chinese), Ministry of Foreign Affairs of the People’s Republic of China, April 30, 2026.

2. Chinese delegation to the 2026 NPT Review Conference, “Reply Statement at the Eleventh Review Conference of the Treaty on the Non-Proliferation of Nuclear Weapons on April 29” [in Chinese], Ministry of Foreign Affairs of the People’s Republic of China, April 30, 2026.

3. The White House, “Fact Sheet: President Donald J. Trump Secures Historic Deals with China, Delivering for American Workers, Farmers, and Industry,” May 17, 2026.

4. State Council Information Office of the People’s Republic of China, “China’s Arms Control, Disarmament, and Nonproliferation in the New Era” (in Chinese), white paper, November 27, 2025.

5. Office of the Secretary of Defense, “Military and Security Developments Involving the Peoples Republic of China 2022” annual report to Congress, U.S. Department of Defense, December 2022.

6. Office of the Secretary of Defense, “Military and Security Developments Involving the Peoples Republic of China 2025,” annual report to Congress, U.S. Department of Defense, December 2025.

7. Vipin Narang and Pranay Vaddi, “How to Survive the New Nuclear Age,” Foreign Affairs, July/August 2025 (June 24, 2025); Andrew F. Krepinevich Jr., “Lighting the Path Ahead: Scenario-Based Planning in the Tripolar Nuclear Age,” Hudson Institute, January 8, 2026.

8. Marco Rubio, “The Next Era of Nuclear Arms Control,” U.S. Department of State, February 6, 2026.

9. Moritz S. Graefrath and Mark A. Raymond, “America’s Allies Should Go Nuclear,” Foreign Affairs, November 19, 2025; Henry Sokolski and Thomas D. Grant, “How to leverage US nukes overseas,” Breaking Defense, April 17, 2025.

10. Dan Plesch and Manuel Galileo, “What the West Misses About China’s Nuclear Build-up,” The Diplomat, March 27, 2026; Frank A. Rose, “Bringing China into the fold on arms control and strategic stability issues,” Brookings Institution, September 25, 2019.

11. Robert Jervis, The Meaning of the Nuclear Revolution: Statecraft and the Prospect of Armageddon (Ithaca, NY: Cornell University Press, 1989).

12. Herman Kahn, On Escalation: Metaphors and Scenarios (New York: Frederick A. Praeger, 1965).

13. Colin S. Gray and Keith Payne, “Victory Is Possible,” Foreign Policy, No. 39 (Summer 1980), pp. 14-27.

14. Gregory Kulacki, “Chickens Talking With Ducks: The U.S.-Chinese Nuclear Dialogue,” Arms Control Today, September 30, 2011.

15. Keir A. Lieber and Daryl G. Press, “The Rise of U.S. Nuclear Primacy,” Foreign Affairs, March/April 2006 (March 1, 2006).

16. William J. Clinton, “The President’s News Conference With President Jiang Zemin of China in Beijing,” The American Presidency Project, University of California, Santa Barbara, June 27, 1998.

17. “United States Experts Call on Japan Not to Oppose a U.S. No-First-Use Policy,” No-First-Use Global, August 10, 2021.

18. Keir A. Lieber and Daryl G. Press, The Myth of the Nuclear Revolution: Power Politics in the Atomic Age (Ithaca, NY: Cornell University Press, 2020); Austin Long and Brendan Rittenhouse Green, “Stalking the Secure Second Strike: Intelligence, Counterforce, and Nuclear Strategy,” Journal of Strategic Studies, Vol. 38, No. 1-2 (2015), pp. 38-73; Brendan R. Green and Austin Long, “The MAD Who Wasn’t There: Soviet Reactions to the Late Cold War Nuclear Balance”, Security Studies, Vol. 26, No. 4 (2017), pp. 606-641.

19. The White House, “President Trump Has Always Been Clear: Iran Cannot Have a Nuclear Weapon,” June 17, 2025; Andrea Shalal and Steve Holland, “Trump says he wants to work with Russia, China on limiting nuclear weapons,” Reuters, February 13, 2025.

July/August 2026
By Douglas B. Shaw, Stephen Herzog, and William C. Potter

“All models are wrong, but some are useful,” observed statistician George E.P. Box.¹ Indeed, today’s large language models and other artificial intelligence systems are far from perfect. Yet, as components of larger social and technical systems, these tools may be reaching a threshold at which their design could shape the future of world order as much as the choices of their users. Frontier AI laboratories have consequently become powerful actors, capable of altering global security faster than states and international organizations can design regulations.² Decisions about whether and how to release advanced models can directly affect risks to humanity posed by weapons of mass destruction (WMD).

In response, the James Martin Center for Nonproliferation Studies launched the “Asilomar process for AI and global security” April 8-9 with a conference titled “Silicon, Swords, and Ploughshares: The Perils and Promise of AI in the Nuclear and Biological Domains.”³ The meeting in Pacific Grove, California, brought together more than 100 experts from universities, think tanks, governments, U.S. national laboratories, and the AI industry. The objective was to begin imagining a new chapter of anticipatory arms control necessitated by the collision of powerful AI tools with catastrophic dangers such as nuclear war, WMD proliferation, and pandemics.

The conference addressed a gaping disconnect between two ways of thinking about AI risks. One, often called AI “doomerism,” views AI as a possible source of human extinction or loss of control over our collective future. The other, a narrower misuse paradigm, asks whether low-capability actors could use AI to build a nuclear weapon, precipitate a meltdown of a nuclear power plant, or engineer a pathogen capable of causing a pandemic. But much practical work lies between these camps. Governments, AI firms, security experts, and the global epistemic community concerned with human survival must jointly shape technology before today’s choices become tomorrow’s constraints.

Why AI Needs Anticipatory Arms Control

Anticipatory arms control seeks restraint before a technology arrives in forms that are already militarized or difficult to restrict. AI fits poorly with the legacy of treaty-based constraints on capabilities that states can define, limit, and monitor. Its most important effects may appear through systems it changes rather than objects that can be counted or banned. In the nuclear and biological domains, AI could reshape proliferation pathways and elements of strategic stability.

This WMD-relevant challenge also arrives as arms control institutions are perhaps least able to absorb a fast-moving technological shock. The New Strategic Arms Reduction Treaty expired February 5 without a replacement. The 2026 nuclear Nonproliferation Treaty (NPT) Review Conference failed to achieve a consensus final document, and negotiators removed all language pertaining to AI from the draft text.⁴ Meanwhile, the Biological Weapons Convention (BWC) still lacks an effective verification mechanism.⁵

Changes in military innovation compound these difficulties.⁶ AI’s most consequential capabilities are emerging from frontier laboratories and startups, often before regulators even understand the problems and opportunities that they pose. The technology rewards speed, offering market incentives to labs and strategic advantages to states.⁷ That uncertainty and private sector acceleration make the current limitations of AI models potentially misleading.⁸ Complacency here would be ill advised, given the technology’s rapid trajectory.

Jane Vaynman and Tristan Volpe’s work on “dual-use deception” helps explain why formal AI arms control may be contentious.⁹ Restraint becomes harder when military and civilian applications are difficult to distinguish and a technology is deeply integrated into military systems and the civilian economy. AI has this character in spades. Threats posed by AI depend far less on the model itself than on its user, what information it can access, and what decisions it is allowed to shape.

The Asilomar process aims to encourage anticipatory arms control before the machinery of governance has caught up. It asks whether laboratories building frontier systems can sit with governments and weapons experts to shape choices that may become facts on the ground. Once model capabilities are released, adopted by nuclear planners, or woven into biological infrastructure, the space for restraint narrows considerably.

Principles for a Missing Forum

The process deliberately invokes a tradition of scientific self-governance. The 1975 Asilomar Conference on Recombinant DNA in Pacific Grove, California, addressed an emerging biological technology whose risks were still being defined.¹⁰ Later gatherings at the same site helped make “Asilomar” synonymous with scientists confronting the social consequences of their work.

The April 2026 meeting tested whether the right communities could reason together before a crisis. The process begins with the premise that AI developers, governments, and security experts each see only part of the picture. There is a missing standing forum where those partial perspectives can be bridged early enough to improve AI governance.

After further deliberation, the conference secretariat released seven principles for governing AI applications in nuclear and biological security, available in full elsewhere:11

• AI must protect human survival.
• Nuclear weapons use decisions must remain under meaningful human control.
• AI governance must strengthen nonproliferation and strategic stability.
• AI developers must contribute to anticipatory risk governance.
• AI-enhanced monitoring and verification must be responsible and ethical.
• AI governance must be globally inclusive.
• AI must not enable disinformation or attacks on nuclear and biological facilities.

Importantly, the principles cover both model-level choices inside AI laboratories and the responsibilities of states, international organizations, and the arms control community. They are meant to give these actors a common starting point to focus on restraint before emergent capabilities harden into crises.

Lessons for Nuclear and Biological Security

The gathering clarified that most AI-related nuclear and biological risks may lie between the poles of extinction speculation and the “bomb in a cave.” The latter phrase captures the fear that a terrorist group or technically weak state could use AI to leap over material and organizational barriers to building a weapon. That concern is legitimate, especially in synthetic biology, but it should not define the field of vision.

Material constraints to building and maintaining nuclear weapons are considerable: fissile materials, precision manufacturing, delivery systems, testing, and command and control. Recent research suggests, however, that AI could eventually help technically advanced would-be proliferators navigate bottlenecks to weaponization.¹² Likewise, purpose-built AI might assist nuclear-armed states in designing more sophisticated warheads. AI integration into command-and-control or decision-support systems may also shape how states in deterrence relationships interpret warnings or assess adversarial behavior.¹³ The late Daniel Ellsberg described multiple large nuclear arsenals ready for prompt use as a single, globe-spanning “doomsday machine.”¹⁴ It might be triggered by any of several governments, but none can prevent this independently. AI could deepen that danger if it makes leaders believe they have found a fleeting advantage or a usable window for escalation. AI also might enable nonstate actors to identify vulnerabilities in civilian nuclear facilities, which could be exploited to unleash nuclear violence.¹⁵

Biological risks, made vivid by COVID-19, raise a different problem. Advanced AI systems may make biological design feel less like specialized knowledge and more like an assisted workflow, especially when paired with cloud laboratories that let remote users run experiments.¹⁶ No one should pretend that engineering a pandemic would be easy. Yet, those responsible for monitoring and verifying such events may increasingly not know where to look until risky work is already well underway.¹⁷

AI also could strengthen tools for evaluating compliance with agreements and norms without replacing human judgment.¹⁸ Machine learning applied to satellite imagery may help analysts notice changes at sensitive facilities.¹⁹ Data fusion across platforms could assist interpretation of noisy indicators, from procurement patterns and orders for biological materials or services to remote laboratory activity. Digital twins of sensitive facilities could help inspectors from the International Atomic Energy Agency and the Organisation for the Prohibition of Chemical Weapons prepare better questions before they arrive at sites.

Although not every catastrophic scenario involving advanced technology belongs in the same category, the lesson for nuclear and biological security is that AI may profoundly alter both landscapes long before the usual guardians of restraint know how the world has changed. The Asilomar process endeavors to look beyond weak actors seeking shortcuts and ask how powerful tools may shape decisions about WMD before new patterns become normalized.

Where Diplomacy Has Not Reached

The process is meant to strengthen, not sidestep, existing forums tasked with ensuring nuclear and biological restraint. The NPT and BWC remain indispensable, but neither is well-positioned to absorb the AI question quickly. Consensus diplomacy can discourage bold and timely action regarding emerging technologies, while the BWC’s absence of verification will become far more consequential as AI touches the practical work of synthetic biology. Concerningly, AI could potentially enable deadly biological capabilities to emerge before anyone can confidently say that a weapons program exists.

AI diplomacy has fortunately moved faster in some spheres than arms control diplomacy, although at a higher level of generality. The AI safety summits at Bletchley Park in 2023, Seoul in 2024, and Paris in 2025 drew international attention to frontier AI risk.²⁰ The “responsible AI in the military domain” process—with summits in The Hague in 2023, Seoul in 2024, and A Coruña in 2026—has also done important work.²¹ These gatherings have a considerable international reach, but their breadth is also limiting. Nuclear and biological arms control can disappear inside larger conversations about AI safety, innovation, military autonomy, and responsible uses.

The Asilomar process is intended to fill that vital space. It focuses on catastrophic nuclear and biological weapons dangers, can move faster than treaty review processes, and aims to bring sometimes disparate expert communities into conversation.

AI’s ‘Pugwash’ Moment?

The April 2026 meeting was only the beginning. The process is an ongoing effort to support new standards of anticipatory arms control among frontier AI laboratories, governments, and security experts. But the next phase must expand the circle beyond participants already convinced that AI will shape global security. Many firms, investors, and research communities driving humanity’s AI future do not yet see nuclear and biological restraint as their responsibility. Many also harbor heroic expectations about the capacity of governments to regulate exquisite tools still being imagined inside the private sector. The time to develop shared AI standards and thresholds in the nuclear and biological domains has already arrived, while oversight is dramatically lagging.

To their credit, some forward-leaning frontier labs have begun hiring staff to think seriously about security consequences of their most powerful models, but the pool of qualified talent is shallow and there is great competition for those individuals. Some firms also have experimented with more restrictive model release practices.22 As a result, several leading AI companies already have endorsed the need for an ongoing Asilomar process, as have multiple states. They share a recognition that the primacy of state governance on issues involving dual-use disruptive technologies has diminished and is unlikely to change trajectory. As such, early engagement between industry and government offers the best prospects for ensuring the development of technologically sound rules and practices that serve both commercial and national security interests.

Even with deeper industry engagement, the participant base must include intergenerational voices from around the world, a point underscored by India’s hosting of the 2026 AI Impact Summit in New Delhi.²³ A process shaped predominantly by powerful Western states and their best-capitalized companies would lack legitimacy and likely reproduce inequalities that have long undercut nuclear and biological governance. Those who bear the consequences of new rules should have a role in shaping them.

A second Asilomar conference is planned for April 2027—by the James Martin Center in partnership with the UN Institute for Disarmament Research and the Berkeley Risk and Security Lab—to deepen and widen this discussion. The process takes inspiration from the Pugwash Conferences on Science and World Affairs, founded in 1957 and later awarded the Nobel Peace Prize for helping scientists pull the world back from the nuclear brink.²⁴ There is reason for cautious optimism that this process can awaken similar awareness and collaboration among those who shape transformational AI technologies.

ENDNOTES

1. George E.P. Box, “Robustness in the Strategy of Scientific Model Building,” in Robustness in Statistics: Proceedings of a Workshop, eds. Robert L. Launer and Graham N. Wilkinson (New York: Academic Press, 1979), p. 202.

2. On this “pacing problem,” see Gary E. Marchant, Braden R. Allenby, and Joseph R. Herkert, eds., The Growing Gap Between Emerging Technologies and Legal-Ethical Oversight: The Pacing Problem (Dordrecht, Netherlands: Springer, 2011); Justin Key Canfil, “Convergent Flexibility: How International Law Keeps Pace with Technological Change,” International Organization, Vol. 80, No. 2 (Spring 2026): pp. 237-280.

3. James Martin Center for Nonproliferation Studies, “CNS Organizes Iconic Asilomar Conference on AI and WMD,” April 14, 2026.

4. See “Implementation of the Treaty on the Non-Proliferation of Nuclear Weapons for its Twelfth Review Cycle,” 2026 Review Conference of the Parties to the Treaty on the Non-Proliferation of Nuclear Weapons, NPT/CONF.2026/CRP.4, May 21, 2026.

5. Matthew P. Shearer, Christina Potter, Rachel A. Vahey, Nancy D. Connell, and Gigi Kwik Gronvall, “BWC assurance: increasing certainty in BWC compliance,” The Nonproliferation Review, Vol. 29, Nos. 1-3 (February-June 2022): pp. 47-75.

6. Stephen Herzog and Dominika Kunertova, “NATO and Emerging Technologies—The Alliance’s Shifting Approach to Military Innovation,” Naval War College Review, Vol. 77, No. 2 (Spring 2024): pp. 47-69.

7. Michael C. Horowitz, “Artificial Intelligence, International Competition, and the Balance of Power,” Texas National Security Review, Vol. 1, No. 3 (May 2018): pp. 36-57.

8. David M. Allison, “Integrating Artificial Intelligence with Nuclear Weapons: A New Paradigm of Irreversible Transformation,” The Nonproliferation Review, Vol. 33, Nos. 1-3 (2026): forthcoming.

9. Jane Vaynman and Tristan A. Volpe, “Dual Use Deception: How Technology Shapes Cooperation in International Relations,” International Organization, Vol. 77, No. 3 (Summer 2023): pp. 599-632.

10. Paul Berg, David Baltimore, Sydney Brenner, Richard O. Roblin III, and Maxine F. Singer, “Summary Statement of the Asilomar Conference on Recombinant DNA Molecules,” Proceedings of the National Academy of Sciences, Vol. 72, No. 6 (June 1975): pp. 1,981-1,984.

11. Stephen Herzog, Allison Berke, Yanliang Pan, William C. Potter, and Douglas B. Shaw, “AI is changing biological and nuclear risks; governance must change accordingly,” Bulletin of the Atomic Scientists, June 18, 2026.

12. David M. Allison and Stephen Herzog, “Artificial Intelligence and Nuclear Weapons Proliferation: The Technological Arms Race for (In)visibility,” Risk Analysis, Vol. 45, No. 11 (November 2025): pp. 3,839-3,859.

13. Herbert S. Lin, “Artificial Intelligence and Nuclear Weapons: A Commonsense Approach to Understanding Costs and Benefits,” Texas National Security Review, Vol. 8, No. 3 (Summer 2025): pp. 98-109; Douglas B. Shaw, Isabelle Williams, Patricia Jaworek, Kevin Park, and Pravin Rajan, “Nuclear Security Implications of AI and Emerging Technologies: A FutureSafe Analysis of Risks and Opportunities,” Nuclear Threat Initiative, November 13, 2025; and Joshua A. Schwartz and Michael C. Horowitz, “Delegating Destruction: Coercive Threats and Automated Nuclear Systems,” International Organization, Vol. 80, No. 1 (Winter 2026): pp. 179-204.

14. Daniel Ellsberg, The Doomsday Machine: Confessions of a Nuclear War Planner (New York: Bloomsbury, 2017).

15. Sarah Case Lackner and Zaheed Kara, “Artificial Intelligence and Nuclear Security Governance: Addressing the Risks of Frontier AI,” Vienna Center for Disarmament and Non-Proliferation, December 2025; Muhammed Ali Alkış, Zoha Naser, and Sarah Tzinieris, “A Fifth Face? The Evolving Threat of Nuclear Terrorism in the Age of Artificial Intelligence,” The Nonproliferation Review, Vol. 33, Nos. 1-3 (2026): forthcoming.

16. Jeffrey Lee, Bria Persaud, Barbara Del Castello, Allison Berke, and Gustavs Zilgalvis, “Documenting Cloud Labs and Examining How Remotely Operated Automated Laboratories Could Enable Bad Actors,” RAND Corporation, PE-A3851-1, April 2025.

17. Doni Bloomfield et al., “AI and biosecurity: The need for governance,” Science, Vol. 385, No. 6,711 (August 22, 2024): pp. 831-833.

18. Jane Vaynman, “Better Monitoring and Better Spying: The Implications of Emerging Technology for Arms Control,” Texas National Security Review, Vol. 4, No. 4 (Fall 2021): pp. 33-56; Francisco Parada, Nathan Martindale, Alisa Reasor, Scott Stewart, and Lindsey Ukishima, “AI for Nuclear Safeguards Verification,” Oak Ridge National Laboratory, ORNL/LTR-2024/3674, October 2024; OPCW temporary working group on AI, “Artificial Intelligence: Report of the Scientific Advisory Board’s Temporary Working Group,” Organisation for the Prohibition of Chemical Weapons, SAB/REP/1/26, March 2026.

19. Yanliang Pan, “Next-Generation OSINT: Machine Learning for Tracking Nuclear and Missile Proliferation,” The Nonproliferation Review, Vol. 33, Nos. 1-3 (2026): forthcoming.

20. Tony Oweke, “The World Is Trying to Govern AI. The UN Wants In.” Council on Foreign Relations, May 29, 2026.

21. Giacomo Persi Paoli and Yasmin Afina, “AI in the Military Domain: A briefing note for States,” United Nations Institute for Disarmament Research, March 2025; Jules Palayer and Laura Bruun, “Artificial intelligence and international peace and security,” in SIPRI Yearbook 2025: Armaments, Disarmament and International Security (Oxford University Press, 2025), pp. 329-346.

22. See, for example, Anthropic Frontier Red Team, “Assessing Claude Mythos Preview’s cybersecurity capabilities,” Anthropic, April 7, 2026.

23. Indian Ministry of Electronics and Information Technology, “India AI Impact Summit 2026,” February 16-20, 2026.

24. John P. Holdren, “Arms Limitation and Peace Building in the Post-Cold-War World,” Nobel lecture on behalf of the Pugwash Conferences on Science and World Affairs, Oslo, December 10, 1995.

July/August 2026

India commissioned its third nuclear-powered ballistic missile submarine (SSBN), the Arihat-class INS Aridhaman.

Indian media reported that sources within India’s defense and security establishment confirmed the April 3 commissioning, although the government has not officially announced it. Defense Minister Rajnath Singh did, however, post in Hindi on social media April 3, that “It’s not words but power, ‘Aridhaman’!”

Possessing three operational SSBNs is a crucial threshold for India because “To ensure round-the-clock operational deployment in the ocean, at least three submarines are required—one on patrol and the other two in maintenance or transit,” Dinakar Peri, a fellow at the Carnegie Endowment for International Peace, wrote April 24 in an analysis.

The INS Aridhaman also has several technological advances that enhance India’s retaliatory abilities as it seeks to defend against China and Pakistan. These include the ability to carry the K-4 submarine-launched ballistic missile, with a 3,500-kilometer range, once that missile completes its testing phase. The ship has eight vertical launch tubes, twice as many as in the first two submarines in its class.

India has already launched the next Arihat-class submarine, which experts expect to be commissioned next year.

According to the Stockholm International Peace Research Institute’s 2026 Yearbook, India may have begun deploying a small number of warheads on one of its SSBNs for peacetime deterrence patrols during 2025. The authors assign “considerable uncertainty” to their assessment that 12 warheads are deployed in this mode. Any coupling of missiles with their launchers during peacetime would be a departure from previous Indian policy.
—MIA CLARKE

July/August 2026
By Kelsey Davenport

Iran and the United States announced a memorandum of understanding to end their war and begin new talks on a nuclear agreement, but negotiators face significant obstacles to a final deal, including a resurgence in fighting.

Pakistani Prime Minister Shehbaz Sharif announced June 14 on the social media platform X that the United States and Iran reached a peace agreement and that, “Both sides have declared the immediate and permanent termination of military operations on all fronts, including in Lebanon.”

The ceasefire broke down two weeks later when the United States blamed Iran for launching drone attacks on two vessels in the Strait of Hormuz and retaliated. Despite the strikes, the U.S. and Iranian negotiating teams conducted a second round of talks on implementing the memorandum in Doha June 30-July 1.

The United States and Israel initiated the 2026 war when they illegally attacked Iran Feb. 28, prompting retaliatory strikes by Iran that included targets in Gulf states. Several days later, Israel began attacking targets in Lebanon that it claims are tied to the Iranian-supported Hezbollah militia.

Pakistan helped broker an initial ceasefire between the United States and Iran April 8 and mediated talks on the memorandum.

Following Sharif’s announcement, U.S. President Donald Trump confirmed the deal in a June 14 post on Truth Social saying “Congratulations to all!” He announced that the Strait of Hormuz, a critical waterway for shipping oil and other commodities that Iran closed by targeting ships, was now open as a result of the memorandum.

Iranian Supreme Leader Ayatollah Mojtaba Khamenei said in a June 18 message shared with Iranian media that he “issued [his] permission” for Iran to sign the memorandum despite having a “different view” of what should be included. He warned that “in-person negotiations that will take place in the future will not mean acceptance of the enemy's position.”

Trump, in Versailles, and Iranian President Masoud Pezeshkian, in Tehran, signed the memorandum June 17. The next day, Pezeshkian called the memorandum an “historic document.” He wrote on X that “peace will be achieved in the shadow of mutual respect.”

The United States released the text of the 14-paragraph memorandum June 17. It said that the United States, Iran, and their allies “declare the immediate and permanent termination of military operations on all fronts, including in Lebanon, and undertake from now on not to initiate any war or military operation against each other, and to refrain from the threat or use of force against each other.”

Specifically on the nuclear program, the memorandum reaffirmed Iran’s commitment to never “procure or develop nuclear weapons.” It also noted that the United States and Iran “agreed to resolve the disposition of stockpiled, enriched material pursuant … with the minimum methodology to be down-blending on site” under International Atomic Energy Agency (IAEA) supervision. The issue of uranium enrichment and other nuclear issues would be negotiated as part of a final deal, according to the memorandum. In return, the United States committed to “terminate all types of sanctions against Iran,” including at the United Nations. It specified that U.S. primary and secondary sanctions on Iran would be included in the sanctions lifting.

The memorandum committed the United States and Iran to negotiate a final nuclear deal “in maximum 60 days extendable with mutual consent.”

The agreement declared the immediate reopening of the Strait of Hormuz and committed Iran and Oman to “define the future administration and maritime services” through that waterway, in discussion with other regional states. Before the war, the Strait of Hormuz was an open waterway, but Iran now seeks to profit from it.

The memorandum includes a commitment to establish a fund to rebuild Iran and requires that the United States issue waivers for Iranian oil sales. The U.S. Treasury Department quickly issued a license for Iran to sell oil and other petrochemical products through Aug. 21.

Although Vice President JD Vance, who was deeply involved in the memorandum negotiations, repeatedly stated that Iran does not receive any unilateral benefits, the upfront sanctions relief and vague nuclear commitments spurred bipartisan criticism of the memorandum as a win for Iran.

Reaching a final agreement in 60 days will be challenging, as negotiators appear far apart on key issues and will likely face opposition.

Although the memorandum references Iran, the United States, and “their allies,” Israeli Prime Minister Benjamin Netanyahu said his country is not bound by the interim deal. He said June 14 that Israeli troops “will remain in the security zone in southern Lebanon for as long as necessary” to protect against Hezbollah attacks. Netanyahu also continues to reiterate that Iran must give up its entire nuclear program as part of any agreement.

Trump wrote June 18 on Truth Social that he expects a “complete Ceasefire on all fronts” and that all states in the Middle East should “maintain their commitment to allowing our negotiations to beautifully unfold.” He has also threatened to resume attacks on Iran if a deal is not reached, in an apparent violation of the ceasefire.

Despite Trump’s threat, U.S. and Iranian negotiating teams met in Lucerne, Switzerland, June 21-22 for the first round of talks after reaching the memorandum.

Pakistan and Qatar mediated the talks and afterwards said that “encouraging progress has been made, including the creation of a mechanism for further technical talks.” A high-level committee will provide political oversight of the process and there will be working groups on nuclear issues, sanctions, monitoring, and dispute resolution, according to a June 22 statement.

After the talks, Iranian Foreign Minister Abbas Araghchi said on X that they “delivered major progress to end Lebanon War.” He described deconfliction in Lebanon as the first “real test” of the process.

The two sides offered different accounts of what transpired on the nuclear issue. In a press conference following the meetings, Vance said that “we made a lot of good progress on other nuclear talks.”

Iranian Foreign Ministry spokesman Esmail Baghaei said that the United States presented its “positions regarding the nuclear issue,” but there was “no discussion of details.” He suggested that nuclear issues also were not a focus of the second round of talks in Doha.

Trump suggested to The New York Times in a June 14 interview that the United States would accept a 15-year suspension on Iran’s uranium enrichment, moving away from his previous demand for a permanent enrichment ban. Iran is open to a suspension but seeks a shorter duration. Although the memorandum suggests that the United States is now open to diluting the highly enriched uranium in Iran, it still appears that Trump wants the blended-down material shipped elsewhere, whereas Iran wants to keep the enriched uranium in the country.

Before the Lucerne talks, Vance suggested that the two sides would discuss a resumption of IAEA inspections. Iran is legally required to implement a safeguards agreement with the IAEA, but it suspended cooperation with the agency and blocked its access to bombed sites after the 12-day U.S.-Israeli war in June 2025. (See ACT, July/August 2025.)

After the meetings ended, Vance said the United States, the IAEA, and Iran will discuss inspections “this week.”

Iranian Deputy Foreign Minister Kazem Gharibabadi suggested the resumption of inspections is not imminent. He said in a June 24 post on X that Iran held “no meeting with [IAEA Director-General Rafael Mariano] Grossi took place in Switzerland, despite his request,” and that there is “no plan for access to the attacked facilities and nuclear materials.” Inspections will be “resolved with the framework of the final agreement,” he said.

Grossi told reporters at a June 24 press conference in Japan that “inspections will indeed take place.” The IAEA and Iran will be “working on the modalities—dates, procedures, places—very soon,” he said.

July/August 2026
By Xiaodon Liang

A key U.S. Senate committee recommended imposing on the military a regulatory framework for autonomous weapons and military artificial intelligence that emphasizes human judgment and ultimate human responsibility, while largely endorsing the Pentagon’s embrace of lethal autonomous weapons and urging that the department “maximize uses” of these emergent technologies.

The Senate Armed Services Committee endorsed the new framework in its draft of the fiscal year 2027 National Defense Authorization Act, which the panel finished marking up June 10.

The sections pertaining to autonomous weapons and military AI would require the department to “ensure personnel exercise appropriate levels of human judgment” and that relevant systems are “designed and employed in a manner that enables commanders and operators to exercise ultimate human responsibility over the use of force.”

The requirement of “appropriate levels of human judgment” is already established in Department of Defense Directive 3000.09, on autonomy in weapons systems, but has lacked a clear definition. (See ACT, March 2023.)

The new framework seeks to elaborate on the principle by specifying that systems must allow for supervision by human operators, include methods for intervention or termination, include fail-safe mechanisms to enable manual control, provide adequate monitoring data to controllers, maintain records of target selection data and logic, and operate in compliance with “international law, rules of engagement, the law of armed conflict, applicable treaties, and Department of Defense policy.”

The proposed legislation would codify the existing review process mandated by the 2023 directive, while also specifying verification, validation, testing, and evaluation criteria. The department’s office of the Director, Operational Test and Evaluation, which has had its proposed budget slashed by two-thirds in this year’s Pentagon budget request, would be charged with ensuring that covered systems function “as anticipated in realistic operational environments against adaptive adversaries and are sufficiently robust to minimize failures.”

The committee text also would establish an incident repository for tracking data on system failures, unintended behavior, and near-miss events.

The language on autonomy and AI would categorically prohibit the use of these technologies for any “decision to initiate the launch or detonation of a nuclear weapon,” a move sought on a bipartisan basis in 2023 by Sen. Ed Markey (D-Mass.), Rep. Ted Lieu (D-Calif.), Rep. Don Beyer (D-Va.) and Rep. Ken Buck (R-Colo.).

The new Senate Armed Services Committee proposal does not, however, include a ban on the use of AI for the targeting of nuclear weapons, as proposed by Sen. Kristen Gillibrand (D-N.Y.) in a stand-alone bill introduced June 2.

Gillibrand’s bill was one of several introduced in the last few weeks seeking to address autonomous weapons and military use of AI by primarily Democratic members of Congress. The others included separate proposals by Sen. Adam Schiff (D-Calif.), Sen. Elissa Slotkin (D-Mich.), and Sen. Mark Kelly (D-Ariz.) in collaboration with Rep. Suhas Subramanyam (D-Va.).

Ultimately, the Senate committee language largely tracks a bill introduced June 8 by Democratic leadership, represented by ranking member Jack Reed (D-R.I.) and his counterpart on the Senate’s defense appropriations subcommittee, Chris Coons (D-Del.). The text, which was accepted by the Republican majority into the base text of the draft of the key policy bill, also incorporates substantial portions of the Kelly-Subramanyam bill.

The Senate committee text, as based on the Coons-Reed bill, is less specific in its protections against domestic surveillance as compared to the other proposals put forward by Democratic senators.

Elsewhere in the draft bill, the committee recommends that the Senate mandate creation of a robotics and autonomous systems command that would be responsible for “force generation, joint training, interoperability, doctrine development, and operational employment through other combatant commands.”

Reflecting bipartisan concerns about the effects of AI adoption, a separate section in the Senate’s draft act incorporates a bill proposed by Kelly and Sen. Tom Cotton (R-Ark.) that would instruct the Pentagon to assess the effect of AI use “on the maintenance and retention of essential warfighter skills.”

Otherwise, the text is largely supportive of Pentagon efforts to expand usage of generative AI and so-called agents, or AI-powered autonomous software.

The new draft legislation for regulation of autonomous weapons systems and military AI faces an uncertain path toward gaining final passage by the full Congress.

The sparer House variant of the defense policy bill, which was approved by the House Armed Services Committee June 5, instructs the Pentagon to update its directive on autonomous systems and military AI while taking into account many of the issues covered in the Senate bill, but provides less specificity and more deference to the Pentagon.

The recent wave of legislative interest in the Senate in autonomous weapons and military AI reflects growing unease over the Trump administration’s revisions to Pentagon policies on the use of AI.

Secretary of Defense Pete Hegseth instructed the department Jan. 9 to adopt a new policy that vendors be required to accept that their AI models may be deployed for “any lawful use.” The order implied that previous departmental restraints would be reassessed.

In February and March, the Pentagon took unprecedented coercive actions against a prominent U.S. technology firm, Anthropic, when the company rejected contract modifications that removed safeguards against use of its AI models in fully autonomous weapons systems and for domestic mass surveillance. (See ACT, April 2026.)

President Donald Trump signed a national security presidential memorandum on AI in the national security enterprise June 5 that, among other measures, tasks the Pentagon to rewrite Directive 3000.09, which was last updated in January 2023.