The atomic bombs dropped on Hiroshima and Nagasaki marked a pivotal moment in modern history. Originally used for destruction, the splitting of the atom defined the first quantum age and shaped the post-war era with the existential threat of “the bomb” reducing cities to ash. Over the past eighty years, this nuclear age has highlighted the dangers of unchecked proliferation, with widespread testing, insider threats, and “loose nukes” serving as both warnings and catalysts for a global system of technology controls and compliance verification. “Trust but verify” became the tagline for US-Soviet/Russia arms control, while a “never trust, just verify” refrain guided international inspectors tasked with verifying that nuclear material and technology for peaceful uses were not diverted to create bombs. Today, uranium is the world’s most regulated material, but the nuclear age—and the verification regime that underpins it—is at a critical turning point as a second quantum age dawns and quantum technology converges with a host of other disruptive technologies.
Converging Disruption
New disruptive technologies are introducing a new stack of decentralized and converging technologies, such as artificial intelligence (AI), additive manufacturing (AM), autonomous drones, and quantum technologies, which are inherently “dual use.” Their applications offer immense benefits in a data-driven world by enhancing data processing, on-demand production, and anomaly detection. However, they can also be used to conceal, obfuscate, and unveil the hidden, presenting a set of risks and potential opportunities for nuclear verification and strategic stability.
Advances in computing power and new algorithms are catapulting AI into a new stratosphere. Since 2023, large language models (LLMs) have increasingly been adopted in the nuclear industry for various applications, from enabling personal assistants and improving efficiency to predictive capabilities and data analysis. At the same time, LLMs, coupled with deep fakes and other synthetic media, make it cheap and easy to alter or create visual, audio, and text content—and manipulate them well. While propaganda and disinformation are nothing new to the nuclear sector, AI is amplifying their speed, scale, and sophistication, turning online rumor mills into vehicles for mass disinformation and blurring the lines between what is real and what is fake. The “black box” nature of AI also makes its influence difficult to trace, opening new pathways for evasion, falsification, and cyberattacks. With less than three percent of AI research focused on safety overall, there is limited discussion among regulators, policymakers, and the tech industry on risk mitigation measures and where, when, and at which points human intervention is most critical. This is particularly worrying when considering ways in which AI could be used for a nuclear power plant’s control room, as well as in command-and-control systems for nuclear weapons and emergency alert systems.
The civilian nuclear sector is also adapting AM (3D printing) techniques to reduce manufacturing costs and produce components difficult to source, such as the pump impeller installed in a Slovenian reactor in 2017, which was printed without the original design drawings. AM also assists the design and development of small, modular, and advanced fission reactors that require alternative supply chains and next-generation materials that can withstand the high temperatures of fusion reactors. While there are challenges in demonstrating the quality and reliability of printed materials, the ability to produce replacement parts and potentially an entire microreactor is driving a series of research activities and demonstrations. The integration of AM with AI is raising new security and proliferation concerns, heightening long-standing concerns in the nuclear sector about counterfeit supply chains. Like AI, additive manufacturing operates across a decentralized supply chain. It can bypass traditional controls on nuclear technology and delivery systems, given that digital design files are vulnerable to cyberattacks and difficult to detect and regulate. Existing export controls apply to AM, but primarily through overlaps with other technologies, leaving critical gaps. As noted by a recent policy memo, AM’s “integration with other disruptive technologies represents one of the most pressing and underappreciated proliferation challenges today.”
AI-enabled targeting and AI-powered drones have also become tools of modern warfare, as seen in the ongoing conflict in Ukraine. Drones are used for intelligence, surveillance, and reconnaissance missions, as well as attack operations. Repeated Russian drone strikes near the occupied Zaporizhzhia nuclear power plant (ZNPP) have damaged reactor structures, cut power, and even hit one of the IAEA’s vehicles. In June, Ukraine’s Operation Spider’s Web delivered a strategic blow to Russia’s cruise missile fleet with inexpensive, widely available delivery vehicles. Their ability to swarm, evade defenses, and potentially deliver a nuclear payload are redefining modern warfare and nuclear risks.
Quantum technologies add a different layer of disruption. Offering powerful new tools to unveil the hidden world, the second quantum revolution is innovating breakthroughs in computing, communications, and sensors. Currently, sensors are the most mature and will enable the ability to “see” through walls, underground, and undersea, thus allowing faster, more precise detection of missile launchers or submarine movements, eroding second-strike capability, and countering the long-held psychology that mass destruction is mutually assured. At the same time, research on quantum computing is advancing towards the anticipated “Q-Day,” when large-scale quantum computers can break current encryption, thereby “breaking the internet” and exposing the security of command-and-control and emergency alert systems. Quantum start-ups are also exploring a range of AI tools that have an accelerating feedback loop between AI and quantum, with AI potentially propelling quantum systems to scale more efficiently.
An Expanding Verification Toolkit
The convergence of technologies also provides a new set of powerful verification tools. AI allows for more efficient analysis of large data sets, making it useful for detecting anomalies in data, flagging unusual patterns, and enhancing predictive analytics for early warning. The IAEA, for example, is increasingly using AI to analyze data from its surveillance system. Additive manufacturing, when properly secured and monitored, can provide detailed digital production records, allowing for traceability and authenticity checks in real time. AM can also support the rapid deployment of bespoke monitoring tools for inspectors in the field.
Quantum sensors may one day offer inspectors ultra-precise, sensitive, and non-intrusive instruments for detecting clandestine activities and verifying dismantlement. For the IAEA, these innovations are essential to keep pace with a nuclear sector shifting toward smaller, faster, and more distributed technologies. These sensors could also possibly enhance the Comprehensive Test Ban Treaty Organisation’s (CTBTO’s) already impressive International Monitoring System, which uses seismic, hydroacoustic, infrasound, and radionuclide technologies, alongside 16 radionuclide laboratories, to detect and verify nuclear tests underground, in the oceans, and in the atmosphere. The IMS has proven its value even in the absence of treaty ratification, successfully identifying North Korea’s nuclear tests and serving as a tsunami warning network for 20 countries, including the U.S. and Russia.
As our cyber and physical worlds merge, nuclear risk is no longer limited to arsenals—it now extends through algorithms, sensors, and code. Future-proofing verification involves creating a “community of practice” across the private and public sectors and international organizations to develop common vocabularies that reflect the realities of converging technologies. “Never trust, just verify” will continue to serve international inspectors well, but the merging of disruptive technologies will require ongoing active assurance of “verify, verify, verify.” This kind of vigilance not only validates data and identifies anomalies but also monitors systems and technological behavior to prevent a future in which “Q-Day” and another Hiroshima could collide.
Verify, Verify, Verify: How Technological Disruption is Redefining Nuclear Risk
By Cindy Vestergaard
Nonproliferation
Eighty years after Hiroshima, the systems built to monitor and contain nuclear threats are confronting a new kind of disruption. Technologies like artificial intelligence, additive manufacturing, and quantum computing are transforming not only how nuclear weapons could be developed or hidden, but also how they can be detected, verified, and controlled. The tools of verification must evolve to meet a world where threats are more decentralized, faster moving, and harder to trace.
The atomic bombs dropped on Hiroshima and Nagasaki marked a pivotal moment in modern history. Originally used for destruction, the splitting of the atom defined the first quantum age and shaped the post-war era with the existential threat of “the bomb” reducing cities to ash. Over the past eighty years, this nuclear age has highlighted the dangers of unchecked proliferation, with widespread testing, insider threats, and “loose nukes” serving as both warnings and catalysts for a global system of technology controls and compliance verification. “Trust but verify” became the tagline for US-Soviet/Russia arms control, while a “never trust, just verify” refrain guided international inspectors tasked with verifying that nuclear material and technology for peaceful uses were not diverted to create bombs. Today, uranium is the world’s most regulated material, but the nuclear age—and the verification regime that underpins it—is at a critical turning point as a second quantum age dawns and quantum technology converges with a host of other disruptive technologies.
Converging Disruption
New disruptive technologies are introducing a new stack of decentralized and converging technologies, such as artificial intelligence (AI), additive manufacturing (AM), autonomous drones, and quantum technologies, which are inherently “dual use.” Their applications offer immense benefits in a data-driven world by enhancing data processing, on-demand production, and anomaly detection. However, they can also be used to conceal, obfuscate, and unveil the hidden, presenting a set of risks and potential opportunities for nuclear verification and strategic stability.
Advances in computing power and new algorithms are catapulting AI into a new stratosphere. Since 2023, large language models (LLMs) have increasingly been adopted in the nuclear industry for various applications, from enabling personal assistants and improving efficiency to predictive capabilities and data analysis. At the same time, LLMs, coupled with deep fakes and other synthetic media, make it cheap and easy to alter or create visual, audio, and text content—and manipulate them well. While propaganda and disinformation are nothing new to the nuclear sector, AI is amplifying their speed, scale, and sophistication, turning online rumor mills into vehicles for mass disinformation and blurring the lines between what is real and what is fake. The “black box” nature of AI also makes its influence difficult to trace, opening new pathways for evasion, falsification, and cyberattacks. With less than three percent of AI research focused on safety overall, there is limited discussion among regulators, policymakers, and the tech industry on risk mitigation measures and where, when, and at which points human intervention is most critical. This is particularly worrying when considering ways in which AI could be used for a nuclear power plant’s control room, as well as in command-and-control systems for nuclear weapons and emergency alert systems.
The civilian nuclear sector is also adapting AM (3D printing) techniques to reduce manufacturing costs and produce components difficult to source, such as the pump impeller installed in a Slovenian reactor in 2017, which was printed without the original design drawings. AM also assists the design and development of small, modular, and advanced fission reactors that require alternative supply chains and next-generation materials that can withstand the high temperatures of fusion reactors. While there are challenges in demonstrating the quality and reliability of printed materials, the ability to produce replacement parts and potentially an entire microreactor is driving a series of research activities and demonstrations. The integration of AM with AI is raising new security and proliferation concerns, heightening long-standing concerns in the nuclear sector about counterfeit supply chains. Like AI, additive manufacturing operates across a decentralized supply chain. It can bypass traditional controls on nuclear technology and delivery systems, given that digital design files are vulnerable to cyberattacks and difficult to detect and regulate. Existing export controls apply to AM, but primarily through overlaps with other technologies, leaving critical gaps. As noted by a recent policy memo, AM’s “integration with other disruptive technologies represents one of the most pressing and underappreciated proliferation challenges today.”
AI-enabled targeting and AI-powered drones have also become tools of modern warfare, as seen in the ongoing conflict in Ukraine. Drones are used for intelligence, surveillance, and reconnaissance missions, as well as attack operations. Repeated Russian drone strikes near the occupied Zaporizhzhia nuclear power plant (ZNPP) have damaged reactor structures, cut power, and even hit one of the IAEA’s vehicles. In June, Ukraine’s Operation Spider’s Web delivered a strategic blow to Russia’s cruise missile fleet with inexpensive, widely available delivery vehicles. Their ability to swarm, evade defenses, and potentially deliver a nuclear payload are redefining modern warfare and nuclear risks.
Quantum technologies add a different layer of disruption. Offering powerful new tools to unveil the hidden world, the second quantum revolution is innovating breakthroughs in computing, communications, and sensors. Currently, sensors are the most mature and will enable the ability to “see” through walls, underground, and undersea, thus allowing faster, more precise detection of missile launchers or submarine movements, eroding second-strike capability, and countering the long-held psychology that mass destruction is mutually assured. At the same time, research on quantum computing is advancing towards the anticipated “Q-Day,” when large-scale quantum computers can break current encryption, thereby “breaking the internet” and exposing the security of command-and-control and emergency alert systems. Quantum start-ups are also exploring a range of AI tools that have an accelerating feedback loop between AI and quantum, with AI potentially propelling quantum systems to scale more efficiently.
An Expanding Verification Toolkit
The convergence of technologies also provides a new set of powerful verification tools. AI allows for more efficient analysis of large data sets, making it useful for detecting anomalies in data, flagging unusual patterns, and enhancing predictive analytics for early warning. The IAEA, for example, is increasingly using AI to analyze data from its surveillance system. Additive manufacturing, when properly secured and monitored, can provide detailed digital production records, allowing for traceability and authenticity checks in real time. AM can also support the rapid deployment of bespoke monitoring tools for inspectors in the field.
Quantum sensors may one day offer inspectors ultra-precise, sensitive, and non-intrusive instruments for detecting clandestine activities and verifying dismantlement. For the IAEA, these innovations are essential to keep pace with a nuclear sector shifting toward smaller, faster, and more distributed technologies. These sensors could also possibly enhance the Comprehensive Test Ban Treaty Organisation’s (CTBTO’s) already impressive International Monitoring System, which uses seismic, hydroacoustic, infrasound, and radionuclide technologies, alongside 16 radionuclide laboratories, to detect and verify nuclear tests underground, in the oceans, and in the atmosphere. The IMS has proven its value even in the absence of treaty ratification, successfully identifying North Korea’s nuclear tests and serving as a tsunami warning network for 20 countries, including the U.S. and Russia.
As our cyber and physical worlds merge, nuclear risk is no longer limited to arsenals—it now extends through algorithms, sensors, and code. Future-proofing verification involves creating a “community of practice” across the private and public sectors and international organizations to develop common vocabularies that reflect the realities of converging technologies. “Never trust, just verify” will continue to serve international inspectors well, but the merging of disruptive technologies will require ongoing active assurance of “verify, verify, verify.” This kind of vigilance not only validates data and identifies anomalies but also monitors systems and technological behavior to prevent a future in which “Q-Day” and another Hiroshima could collide.
Recent & Related