Friday marks 25 years since the September 11 terror attacks. Americans are spending the week remembering those who lost their lives and honoring the first responders and others who sprang into action to help. The attacks cast a long shadow, stretching beyond Ground Zero, the Pentagon, and Flight 93 to transform not just the United States but the world.
RAND experts are reflecting on one of the most enduring aspects of this transformation: what we've learned about homeland security in the quarter-century since the attacks. Below, our researchers answer questions about protecting soft targets, mitigating risks from emerging technologies, and more.
What has stayed with you most about September 11, 2001? How did that day shape the way you think about protecting the homeland?
Heather Williams My professional trajectory has been shaped by many events, but it was set in motion by September 11. The attacks led me to focus on the Middle East and to join the U.S. Intelligence Community. That day stays with me as a reminder that solidarity is interwoven into the American fabric. This is comforting as we grapple with the current national divisions.
My thoughts on September 11 as a policy expert are more nuanced. Policymaking is so difficult because bureaucracy is complex and people are naturally resistant to change. September 11 allowed leaders to overcome much of the resistance, and I lived the positive effects of the Intelligence Reform and Terrorism Prevention Act. At the same time, the reactionary nature of those changes created a patchwork of organizations working on homeland security, largely focused on terrorism as the threat.
That enterprise is now reorienting toward today's multi-hazard environment but without the catalyst that September 11 provided. Like many others, I'm motivated to build the necessary protective institutions before another tragedy strikes.
Ryan Consaul I think back to the tremendous sense of patriotism and unity born from the great tragedy of 9/11. Sadly, that unity has given way to controversy and division. Congressional oversight remains diffuse, and past bipartisan efforts to enact meaningful legislation to protect the homeland have given way to more-partisan priorities. Given the multitude of threats we continue to face, we need to stand together to tackle ongoing and emergent homeland security challenges.
Given the multitude of threats we continue to face, we need to stand together to tackle ongoing and emergent homeland security challenges.
What have we learned in the years since 9/11 about how best to protect soft targets? What does the evidence say about safeguarding communities from mass attacks by nonstate actors and others?
Brian Jackson For protecting soft targets, we often think first about technology: cameras, weapon-detection systems, or other physical protection measures. But some of the most effective protection comes from the community working together to identify and respond to potential threats. This is called behavioral threat assessment and management, or BTAM, and it's now one of the ways that the majority of K–12 schools, many workplaces, and some police departments or community safety efforts protect themselves. Teams respond when someone's behavior suggests they may become violent. Much of what the teams using this approach do doesn't look like security. For example, they might get the people referred to them into counseling or other programs to redirect them from whatever is moving them toward violence.
BTAM is effective, but it relies on everyone buying in. You can't intervene with a kid who's reading violent websites or making threats if their classmate doesn't let the teacher know what's happening or if the kid's parent refuses to let the child work with a counselor. Building and maintaining trust is tough. We tried to use this approach in the years after 9/11; it was a central part of efforts to respond to radicalization to violent extremism. But because many of the other counterterrorism approaches undermined trust with key communities, the strategy was never able to strengthen homeland security as much as it could have.
John Hollywood The evidence supports starting with layered security strategies in which multiple measures work together to improve the chance that an attack will be stopped or at least mitigated.
These security layers start with prevention. Tips from the public to police or on-site security have often helped to foil plots. Look for warning signs that incorporate a strong motivation to attack with concrete actions to prepare. Things like creating attack plans; researching how to attack; seeking to learn from extremists (including traveling to receive paramilitary training); accumulating large quantities of weapons or materials used in the construction of improvised explosives; and probing or breaching potential attack sites. Be aware that most motivations for mass attacks have been personal rather than ideological or partisan.
For on-site security, focus on the basics. Cases where attackers had direct access to a large crowd have tended to have the highest lethality, so consider how to put distance, barriers, and crowd movement between would-be attackers and crowds. Access-control systems—notably exterior and interior door locks, secured windows, and securable entryways—have all been effective and efficient. However, all require training and maintenance.
Be aware that most motivations for mass attacks have been personal rather than ideological or partisan.
Bystanders and security have stopped attacks. Groups of bystanders tackling shooters has been extremely effective. Again, training can make responses even more effective. That said, training needs to be low-stress and empowering without causing psychological injuries.
For communities, use interagency teams to support prevention, security training, and response planning. We need to establish and fund teams responsible for both educating the public about what to report and conducting diligent follow-up on reported cases. (Brian mentioned an approach for doing so, BTAM, above.) We then need to ensure advance planning and joint training with those agencies that will respond locally to a mass attack.
For many Americans, the creation of the Transportation Security Administration (TSA) and changes to airport screening remain among the most visible legacies of 9/11. What has research found about the TSA?
Kelly Klima Research over the past 25 years suggests that TSA has played a vital role in reducing the risk of terrorist attacks against U.S. aviation. Since 9/11, there has been no successful terrorist attack on a U.S. commercial flight that passed through TSA screening. This record reflects a layered defense-in-depth approach to screening passengers, baggage, and cargo, alongside intelligence, law-enforcement, and industry partners.
TSA has also balanced security with the need to move millions of travelers and large volumes of cargo efficiently and at reasonable cost. The passenger fee is less than $6 per flight, and the total TSA budget (which includes the development and procurement of screening technology) is about $8 billion in 2026.
That balance of security and reasonable cost continues to improve through better physical and chemical screening technologies, changes in checkpoint design and procedures, and risk-based approaches that focus attention where it's needed most.
But the threat continues to evolve. Terrorists and insiders may seek to conceal firearms, explosives, or other weapons in new ways. Emerging cyber and other nontraditional threats require constant assessment. Continued investment in advanced detection technologies, process redesign, and responsibly developed AI can help TSA identify threats more accurately and quickly, strengthen aviation security, and improve the passenger experience.
Continued investment in advanced detection technologies, process redesign, and responsibly developed AI can help TSA identify threats more accurately and quickly.
Williams Americans associate TSA with the blue shirts they encounter at the security checkpoint, but the organization has a broader mandate than many realize. One of the other programs TSA manages is the Security Threat Assessment for multiple security credentials, including the Transportation Worker Identification Credential (TWIC) and Hazardous Materials Endorsement for commercial driver's licenses. RAND research has found that TSA generally executes these programs effectively, at cost, and with timely resolution, in a way that reduces internal threats. For example, TWIC applicants who have no flags for possible disqualifying factors could find a physical card in their mailbox less than six days later.
Many anti-terrorism technologies and capabilities require collaborations between government and the private sector. What have we learned about how to effectively navigate those partnerships?
Thao Liz Nguyen One of the enduring lessons is that the development, maturation, and operational adoption of anti-terrorism technologies can proliferate in well-structured public-private collaboration. After 9/11, the Support Anti-terrorism by Fostering Effective Technologies Act of 2002, enacted as part of the Homeland Security Act of 2002, established the Office of SAFETY Act Implementation and created a statutory framework for evaluating and incentivizing anti-terrorism technologies. From a technology-evaluation perspective, the SAFETY Act not only encourages innovation in the abstract but applies a rigorous process to assessing technical performance, operational utility, and real-world deployment considerations of mature technologies to emerging capabilities.
Designation and Certification protections under the SAFETY Act have helped mature the market for proven security technologies and services by reducing liability uncertainty and signaling that a capability has undergone meaningful DHS review. At the same time, the Developmental Testing and Evaluation Designation protection has been especially important for emerging technologies, because it allows developers to generate operational evidence, refine concepts of use, and test performance in relevant environments before a capability is fully fielded.
What we've learned is that effective public-private partnership requires more than funding or policy encouragement; it requires credible evaluation pathways, clear incentives for participation, and mechanisms that help bridge the gap between innovation, validation, and scalable adoption for anti-terrorism technologies.
The development, maturation, and operational adoption of anti-terrorism technologies can proliferate in well-structured public-private collaboration.
Of course, new technologies have also transformed the threat landscape. Let's start with unmanned aerial systems, commonly known as drones. How have drones emerged as a potential threat to people and infrastructure, and what do we know about how to mitigate that threat?
Brendan Toland Drones have rapidly emerged as a threat because these systems have become more capable and more available. Parallel advances in a series of technologies (batteries, electric motors, precision manufacturing, optics) have shrunk the airframe for drones, increased their speed and endurance, and lowered their cost. Additionally, improvements in lightweight cameras, guidance via GPS waypoints, and wire-guided control have made it more difficult to detect and take over hostile drones. Finally, drones come in various forms. This lets bad actors select a drone that is optimal for the intended objective, whether that's maintaining an eye in the sky or delivering compact items (contraband, bombs) virtually anywhere at speed.
As part of RAND's work operating the Homeland Security Operational Analysis Center, we evaluated the capabilities of commercially available drones. We found that these systems continue to lower in cost while improving on operationally relevant metrics, such as speed, flight time, and payload capacity. We also developed several high-threat scenarios in which bad actors could use drones to threaten the homeland, including bombing a federal building or delivering a chemical agent outside a major sporting event. Looking across the market, we found more than 500 systems capable of conducting at least one of our high-threat scenarios. This number has only grown in the six years since our research was conducted.
The speed of drones and the ability to launch these systems from virtually anywhere constrains the response time for those working to defend the homeland. In our more recent simulation modeling, we found that defenders had only seconds to a few minutes to detect and defeat incoming drones, depending on the situation. If the adversary uses a drone to provide aerial surveillance, then in many cases the onboard camera enables a stand-off range that undermines detection. Any effective defense requires deploying layered mitigation capabilities in advance, putting these capabilities in the hands of trained personnel, sharing detection information across the homeland security enterprise, and empowering defenders on the ground to make quick decisions.
Christopher Scott Adams Advances in drone technology have made the threat more salient, but the homeland security enterprise has some tools to help mitigate that threat. The Preventing Emerging Threats Act and the SAFER SKIES Act have given key counter-drone authorities to federal and state/local/tribal/territorial law enforcement, respectively. These include permissions, under certain circumstances, to intercept communications between drones and their operators, track and monitor drone flights, and disable, down, or destroy potential threats.
These authorities have allowed DHS and its partners to invest in counter-drone technologies. At RAND, we've advised our DHS sponsors on the requirements for effective counter-drone capabilities and the right technology mix. This analysis has informed ongoing counter-drone investments at the border and in defense of major events, such as the World Cup and the Olympics.
Advances in drone technology have made the threat more salient, but the homeland security enterprise has some tools to help mitigate that threat.
Across these assessments, we've found a few consistent effective practices to mitigate, though not eliminate, the threat. First, defense-in-depth is key; having multiple types of sensors and effectors (kinetic and non-kinetic) in the same location gives the best chance of detecting and defeating a variety of drone threats. Second, defenders need to plan ahead. This includes resolving authorities issues, developing rules of engagement that are well understood by all operators, and creating information-sharing agreements across responsible agencies. Third, employing automation in defense can provide the speed to respond to a similarly automated threat.
What about artificial intelligence? For example, concerns are rising that AI could enable biological attacks. What do we know about this emerging risk and how best to mitigate it?
Steph Guerra Historically, biological attacks have been rare, unsuccessful, and difficult to execute. That's because of the technical, operational, and motivational barriers that bad actors face. But AI is increasingly able to lower these barriers by democratizing biological expertise, driving laboratory workflows, raising the ceiling of biological novelty, and circumventing existing biosecurity controls like gene synthesis screening. At RAND, we seek to assess these capabilities to help inform policy action that reduces the risk of misuse while still enabling beneficial scientific innovation.
Our team recently published a biosecurity strategy for the AI era to prevent AI-enabled biological attacks. The strategy is deliberately layered, because different mitigations stop different threat actors. For example, access controls like gene synthesis screening may stop a lone terrorist, but they probably can't stop a state bioweapons program. But a state program might be deterred from developing bioweapons by credible attribution and rapid response capabilities. Each layer is critical, but only having one is insufficient.
We know AI capabilities in biology are advancing fast, but thankfully the most hazardous thresholds haven't yet been crossed. That gives us a narrow window to build protective infrastructure before misuse becomes demonstrably easy.
Nguyen Detection technology is being pushed to do more than it was originally built for. This is true for any emerging unforeseen threat. As AI lowers the barrier to entry of designing novel biological agents or even modifying existing ones, screening systems must improve at spotting unusual patterns, not just known signatures stored in their static reference libraries. In practice, that means detection equipment needs to keep pace with new compounds, new delivery methods, and changes in how material may be packaged or concealed.
The challenge is that detection is only as good as the data, calibration, and reference libraries it relies on, so if the libraries are outdated, then the equipment can miss something that doesn't look like the “expected” threat. That's why DHS and other stakeholders should continue investing in research, development, testing, and evaluation; field testing; and recalibration of sensors and screening tools. We also need to think about how these systems are integrated across layers of screening, because no single device is going to catch everything. We need to invest in innovative detection systems and the algorithms behind them so they can stay adaptive and agile as threats evolve.
David Luckey The convergence of AI with advances in biotechnology and chemistry represents one of the most significant emerging risks to homeland security since 9/11. AI is rapidly lowering technical barriers in synthetic biology, genomics, and chemical engineering. This means that malicious actors could potentially leverage AI tools to design, synthesize, or optimize biological or chemical agents with unprecedented speed and precision.
This risk is not hypothetical. AI-driven platforms can already assist in protein folding, gene editing, and even the design of novel chemical compounds. Used by a bad actor, these capabilities could be repurposed to engineer pathogens or toxins that evade current detection and mitigation strategies. The dual-use nature of AI in life sciences poses unique governance and oversight challenges; the same tools that accelerate medical breakthroughs can also facilitate weaponization.
Mitigating this risk almost certainly requires a multi-layered approach that includes strengthening oversight, promoting responsible innovation, enhancing detection and response, and building cross-sector partnerships. Effective mitigation will likely depend on seamless collaboration between government, industry, and the research community. Proactive risk assessment, horizon scanning, and anticipatory governance are essential to stay ahead of adversaries who might exploit AI for catastrophic ends.
Proactive risk assessment, horizon scanning, and anticipatory governance are essential to stay ahead of adversaries who might exploit AI for catastrophic ends.
Twenty-five years later, what lesson from 9/11 feels most urgent in your work today?
Jackson Responding to threats requires thinking about how we will maintain defenses in the long term, even when the need to respond is most urgent.
The human and financial costs of 9/11 were catastrophic, and so there was pressure to do everything possible to make sure that such a tragedy couldn't happen again. As time passed, it became clear that the costs of all the security and other measures added up over time—and that one of the most effective responses to that specific threat (reinforcing airplane cockpit doors) was one of the cheapest. While government needs to spend to protect the nation, the more expensive security measures are, the harder they are to maintain over time.
Though thinking about the cost-effectiveness of security may seem odd when we're concerned about the risk of disasters like 9/11, doing so is critical, particularly since it's often much cheaper for attackers to try something new than it is for defenders to update protections.
Hollywood We need to stay vigilant but not overreactive.
This means maintaining a focus between overreaction to what have been extremely rare events and inattention to ongoing threats. Similarly, we must avoid getting distracted by a single part of the threat (e.g., focusing on a specific ideology when most attacks are personally motivated) or by a single security technology. Instead, what's needed is a strategic, layered approach that counters the full range of violent threats. Perhaps most importantly, vigilance means maintaining and developing partnerships between the public, security groups, and local, state, and federal agencies to provide ongoing prevention and protection.
Luckey It's imperative to have imagination and adaptability in the face of evolving threats.
9/11 exposed the dangers of “failure of imagination,” specifically the inability to anticipate how adversaries might exploit emerging or unexplored vulnerabilities. Today, as we confront the intersection of AI and biological or chemical weapons, that lesson is more urgent than ever.
Our adversaries are agile, opportunistic, and increasingly technologically sophisticated. The tools of mass disruption are no longer confined to nation-states or well-resourced groups; they're becoming accessible to a broader spectrum of actors. This reality demands that we continually reassess our assumptions, update our risk models, and foster a culture of innovation and vigilance across the homeland security enterprise.
I'm reminded daily that security is a dynamic process, not a static end state. The post-9/11 era amplified the value of intelligence sharing, public-private partnerships, and resilience. But it also illuminated that the next threat might not look like the last. Our greatest asset is the ability to anticipate, adapt, and act—before the unimaginable becomes reality.
