Understanding and Managing Global Catastrophic Risk

Image by SasinParaksa/Getty Images; elements of image from NASA

The world faces risks from natural hazards and human-caused threats that could significantly harm or set back human civilization at the global scale (i.e., catastrophic risk) or even result in human extinction (i.e., existential risk). Some threats, such as severe pandemics, bring the potential for massive numbers of fatalities in a short time. Nuclear war could do the same while also destroying infrastructure, economies, and the function of national governments. Some hazards, such as climate change or supervolcanoes, have the potential to disrupt the natural environment and ecosystems in ways that threaten the stability of society and human health and welfare. Existing and potentially emerging advances in AI could erode the foundations of human capabilities. Extreme versions of any of these hazards or threats could simultaneously introduce all these effects.

In 2022, Congress passed the GCRMA to provide policymakers, emergency management planners, and other stakeholders with a strategy to respond to catastrophic risks. The first requirement of the legislation is for the Secretary of Homeland Security and the Federal Emergency Management Agency (FEMA) administrator to produce a comprehensive assessment of global catastrophic and existential risks associated with

• asteroid and comet impacts
• supervolcanoes
• severe pandemics
• rapid and severe climate change
• nuclear conflict
• AI.

In response to a request from the U.S. Department of Homeland Security (DHS) — specifically, FEMA — for support in meeting the GCRMA's assessment requirement, RAND researchers developed a risk summary for each of the hazards and threats noted in the law.

How Did the Researchers Assess Risk?

In the risk management field, risk is often defined as the combination of a hazard or threat event's consequences and likelihood (see Figure 1). The risk assessments developed by the RAND team examine how policymakers and stakeholders can (1) understand catastrophic risks — by identifying how a hazard or threat leads to such risk, estimating the likelihood and potential consequences of each risk, and highlighting the uncertainties associated with such efforts — and (2) manage catastrophic risks — by estimating how each risk is likely to change in the next decade and identifying known and potential mitigation options.

The hazard is a supervolcano. The hazard event is a supereruption. The likelihood of a supereruption has annual exceedance probability: 6.7 × 10^–5, with an approximate minimum return period of 15,000 years. The consequence is damage to the natural environment and economic stability, which, in turn, can cause societal instability, death, and reduced human capability. The likelihood and consequence compose the risk.

Understanding Global Catastrophic Risk

Hazard and Threat Consequences Can Result in a Variety of Outcomes

Each assessment began with an examination of the natural and human processes that combine to create that hazard or threat's potential consequences in one of the following four categories:

• mortality, a rough number (order of magnitude) of human fatalities that might be expected
• ecosystem instability, disruption to the natural environment and ecosystem functions on which humans depend
• societal instability, disruptions to economic and governance systems
• reduced human capability, degradation in key dimensions of human well-being.

Together, these four consequence categories accommodate consideration of the spectrum of factors that influence how people judge and manage risks.

As events that threaten civilization unfold, over time, society would likely experience consequences in all four categories of risk. That said, the authors focused the risk assessments on the direct effects of the hazard or threat in question, acknowledging that indirect effects also merit consideration.

Table 1 presents the most-significant consequences identified for each of the six hazards and threats in their respective assessments.

Grappling with Probability and Uncertainty

Those responsible for emergency management might encounter significant uncertainty when developing strategies to respond to global catastrophic and existential risks. Extreme events are infrequent and, in some cases, have little or no precedent. Some hazards are well understood scientifically, whereas others remain deeply uncertain in terms of how they might occur, how effects might evolve, or both. Planners face even greater ambiguity about how and when emerging technologies, such as AI or synthetic biology, will transform the threat landscape or, alternatively, offer solutions to global challenges.

Efforts to estimate the probability of global catastrophe and human extinction must navigate all these forms of uncertainty. In some cases, probability estimates can be made, such as those shown in Table 2. In others, only factors driving likelihood and potential scenarios can be described. And, for the most-uncertain risks, even the potential scenarios leading to catastrophe defy characterization.

Variation in the Geographic Extent of Hazard and Threat Consequences

Some hazard and threat consequences that the researchers assessed — such as the blast and thermal effects from the impact of a small asteroid or the effects of hurricanes made more intense or more frequent by climate change — would likely have regionalized consequences that would not be expected to overwhelm the capacity of a geographically larger country to respond and recover. For these types of events, countries develop plans to mitigate damage, warn populations, respond to incidents, and aid neighboring countries or regions in recovery should disaster strike.

As the geographic extent of hazard and threat consequences expands, so does the scope of events and considerations of national concern. Nuclear conflict anywhere in the world could destabilize economies and heighten conflicts that threaten friends and allies. A disease outbreak or bioterrorism, even if contained or involving a noninfectious agent, demands responses that affect trade, travel, and migration. Climate change–enhanced natural disasters could overwhelm some countries' responses and increase calls for foreign assistance and aid for recovery. For hazards and threats like these, with considerations that extend more broadly, countries plan both to contain the cascading consequences of events when they occur and to fulfill global responsibilities to help others and address risks when they do spread globally.

Events of the largest scale described in the report have global consequences or effects that are not contained geographically. Asteroids larger than 1,000 m in diameter, the largest supervolcanoes, runaway climate change, and global nuclear war hold the potential to affect all countries and all people. Advances in AI, although poorly defined and understood, have the potential to transform human civilization globally.

Factors That Will Affect Levels of Risk in the Future

Trends Affecting Individual Global Catastrophic Risks

Overall, global catastrophic risk has been increasing in recent years and appears likely to increase in the coming decade. For two hazards — supervolcanoes and asteroid and comet strikes — risk will likely remain constant or reduce in the next decade, in large part thanks to improved detection and response capabilities. For three hazards and threats — climate change, pandemic, and nuclear war — the risk appears to be increasing in the next decade because of current or expected human activities. For AI, the uncertainties are sufficiently large that it is difficult to determine anticipated changes in risk with any confidence. Figure 2 provides more detail on the reasons individual hazard- and threat-related risk might be increasing or decreasing.

Changes in Risk Levels in the next 10 years

Threat or Hazard: AI = increase

Threat or Hazard: Asteroid or comet impact = stays the same or decrease

Threat or Hazard: Nuclear war = increase

Threat or Hazard: Climate change = increase

Threat or Hazard: Pandemic = increase

Threat or Hazard: Supervolcano = stays the same

NOTE: More information about the reasons the total risk for each hazard and threat might increase or decrease in the next decade is included in the report.

Common Factors Affecting All Six Sources of Global Catastrophic Risk

Several common factors influence global catastrophic risk from all six threats and hazards assessed in the RAND report:

• the rate and nature of technological change
• the maturity of global governance and coordination
• failure to advance human development
• interactions among the hazards themselves.

How these drivers evolve can determine how, how seriously, and how swiftly the United States and countries worldwide will need to respond to manage the risks. Plans that reflect these drivers will be more robust to the uncertainties they create.

Managing Global Catastrophic Risk

Aligning Risk Management to the Nature of Specific Hazards

As noted above, the hazards and threats reviewed by RAND can vary widely in terms of the geographic extent over which consequences can be expected to occur and the quality of understanding about their scope, likelihood, and consequences. These two dimensions of risk, which are plotted in Figure 3, influence the appropriateness of risk management approaches.

The y axis -- titled "Quality of evidence to support risk management" -- starts at the bottom at "High" and ends at the top at "Low."

The x axis -- Geographic extent of consequences" -- starts at the left at "Regionalized Global" and ends at the right at "Regionalized Global."

The following items appear as bubbles in the chart with the first items at the top in the "low" category" and last near the bottom in the "high" category.

• AI's effects on existing risks
• Intentional pandemics, including those enhanced with synthetic biology
• Unaligned artificial general intelligence
• Exceedance of thresholds for key global climate risks
• Global nuclear war
• Comets
• Global atmospheric effects of a supervolcano
• Small asteroids
• Caldera eruption (local effects)
• Medium-sized asteroids
• Natural pandemics
• Climate change-enhanced natural hazards
• Large asteroids

NOTE: The placement and sizes of the ovals in this figure represent a qualitative depiction of the relative relationships among threats and hazards based on interpretation of aspects of the assessments described in the report. The figure presents only examples of some cases and scenarios described. Artificial general intelligence generally refers to AI systems that can do at least as well as humans can on cognitive tasks, handle unanticipated problems, and generalize what they learn. A caldera is a large crater caused by certain kinds of volcanic eruptions. For more about these concepts, see the main report.

For the threats and hazards with relatively well-understood effects and responses and where the geographic scale of the consequences aligns with the appropriate risk management jurisdictions (i.e., those in the lower left-hand corner of Figure 3), FEMA, DHS more generally, and other parts of the U.S. government could begin to organize useful responses. For example, for small asteroids, the government could improve the National Aeronautics and Space Administration's (NASA's) planetary defense capabilities and improve capabilities for evacuation and civil defense. Such evacuation and civil defense capabilities would also prove useful for the local effects of a supervolcano and for a limited nuclear attack. Other events of this scale and relative understanding can leverage existing planning frameworks, such as the National Preparedness System and conventional approaches to planning for continuity of government and continuity of operations.

Addressing the risks in the other corners of Figure 3, however, will require significant innovation to generate (1) the capacity for currently unknown or unavailable responses, (2) risk management approaches suitable for such deeply uncertain risks, and (3) enhanced institutions at all levels of governance (including internationally) able to implement these risk management approaches. Such innovation could also enhance risk reduction for the better-understood risks in the lower left-hand corner of the figure.

The geographic scale of the consequences of a risk is not the same as the geographic scale of the most-appropriate responses to a risk. Many of the risks shown in Figure 3 might be best addressed with a coordinated global response to most effectively prevent or address the consequences regardless of where disaster strikes.

Risk Management Interventions

Because a global catastrophic risk management strategy may include many different types of actions, thinking broadly about opportunities to intervene to manage risks is necessary (see Figure 4). For example, reduction or prevention of the onset of a threat or hazard might be possible. When neither of those is possible or successful, opportunities could arise to disrupt the mechanisms that lead to harmful consequences. Furthermore, there might be ways to reduce the severity of effects should they occur. And, when all else fails, options exist to recover from the effects.

Preventing Impact

• Reduce the onset of the threat or hazard
• Disrupt mechanism leading to risk

Reducing the Consequences of Impact

• Reduce severity of effects
• Enhance response and recovery

Preventing Impact

As of early 2024, risk mitigation strategies targeted to prevent catastrophic impact from the six assessed hazards and threats spanned stages of development and capability. Approaches for the least-mature cases are theoretical and untested (e.g., draining heat from the magma chamber of a supervolcano or in-space deflection of large impactor asteroids and comets). Also, some risk management approaches might be well studied, tested, and more mature but depend on controlling human behaviors, such as curtailing greenhouse gas emissions or preventing nuclear war. In the case of AI, both current and future threats remain uncertain. Nevertheless, governmental and nongovernmental organizations have produced frameworks to help prevent the deployment of systems that could cause harm, and the AI community has proposed risk management strategies, such as international technical standards, regulations, voluntary self-regulation, and market pressures.

Reducing Consequences of Impact

Strategies to reduce the severity of hazard or threat impact also vary but can closely align with existing emergency preparedness frameworks when considering response and recovery options. Even so, the most-salient approaches should be determined by characteristics of the risks posed by each hazard and threat. For example, reducing the consequences of climate change or nuclear war involves building resilience in affected or potentially affected communities through such actions as relocating at-risk populations or building shelters. Preparing for a supervolcano or a large asteroid impact can take a similar approach through evacuation planning. These approaches will require significant investment in governments' and communities' research and development of early-warning systems and preparedness planning. With respect to reducing the severity of pandemics, recent examples of vaccine hesitation and mistrust among the public show that government institutions will need to improve strategic communication and work to build trust with communities, in addition to pursuing accelerated vaccine development and distribution.

	AI	Asteroid and Comet Impact	Nuclear War	Climate Change	Severe Pandemic	Super-volcanoes
Reduce the likelihood of occurrence.	Governance and policy Research and development	Research and development	Governance and policy Technical and logistical	Governance and policy Technical and logistical Research and development	Governance and policy Technical and logistical Research and development	Research and development
Disrupt the mechanisms leading to the risk.	Governance and policy Research and development	Research and development	Governance and policy Technical and logistical	Research and development	Governance and policy Technical and logistical Research and development	Research and development
Reduce the severity of the effects.	Governance and policy Research and development	Governance and policy Technical and logistical Research and development	Governance and policy Technical and logistical	Governance and policy Technical and logistical	Governance and policy Technical and logistical Research and development	Governance and policy Technical and logistical Research and development
Enhance response and recovery.	Not applicable	Governance and policy Technical and logistical	Governance and policy Technical and logistical	Governance and policy Technical and logistical	Governance and policy Technical and logistical Research and development	Governance and policy Technical and logistical

Three Types of Risk Mitigation Efforts

As presented in Figure 5, investments in risk management options can be grouped by the following three categories:

• taking action: Many risk mitigation efforts identified in the researchers' assessments involve technical or logistical measures that can be implemented to prevent or reduce risks.
• governing: Some of the identified risk mitigation options involve using regulations and laws, policies, or norms to influence actions and behaviors in ways that reduce risks.
• learning: For some hazards and threats, there might not be options or enough knowledge about risks to either act or govern, and research and development are necessary before more-active steps can be taken.

Technical and logistical approaches are most relevant for those threats and hazards that are better understood and for which solutions are readily available, such as managing the effects of nuclear detonations, natural hazards exacerbated by climate change, and efforts involving public warnings, evacuations, and incident response and recovery. Although such solutions can directly reduce risk, they can also be costly and require attention to planning and evaluation to ensure effective implementation.

Governance and policy approaches are relevant where risks result from human behaviors and economic activity. For example, regulations and policies can be established to prevent or minimize the misuse or development of dangerous AI and biotechnology or to reduce emissions of carbon into the atmosphere. Governance approaches also include coordinated national or global efforts to warn populations of hazards and threats and to respond or support recovery should incidents occur. Like technical and logistical approaches, governance and policy can reduce risks directly. However, governance and policy depend on legal and political systems for successful implementation.

Research and development present opportunities to identify new approaches to manage risks to human civilization and existence. For example, although the ways in which increasing concentrations in atmospheric carbon pose harms are understood, approaches to sequestering atmospheric carbon are only theorized or experimental. Advances in biotechnology and AI hold the potential to create new vaccines and medical countermeasures that are more effective and developed on shorter timelines.

Recommendations for U.S. and Global Policymakers

The six risk assessments conducted by the RAND researchers paint a complex picture of global catastrophic and human existential risk that demands a risk management approach that matches the nature of these threats and hazards, the uncertainties that surround them, and the interactions among them. To do so, the United States and other countries should consider taking several steps toward improving assessment and management of threats and hazards that present the potential for global catastrophic and existential risks:

• Incorporate comprehensive risk assessments into management of global catastrophic and existential risks.
• Develop a coordinated and expanded federally funded research agenda to reduce the uncertainty about global catastrophic and existential risks and to improve the capabilities to manage such risks.
• Develop plans and strategies when global catastrophic and existential risk assessments are supported with adequate evidence.
• Expand international dialogue and collaboration that address global catastrophic and existential risks.
• Adapt planning and strategy development to address irresolvable uncertainties about global catastrophic and existential risks.

These actions should be viewed as preludes to an ongoing quest to understand and manage risks that can threaten humanity. Building and sustaining a resilient world will require continually identifying, assessing, managing, and monitoring the sorts of risks discussed in the report and others that might be lurking in the shadows of society's collective ignorance.

Available for Download

Download PDF

Topics

• Artificial Intelligence
• Emergency Preparedness
• Global Climate Change
• Global Health
• Global Security
• Natural Hazards
• Nuclear Weapons and Warfare
• Public Health Preparedness
• Space Science and Technology