In Transportation Security, Speed Wins - HSToday

In more than 20 years in aviation security, I have never questioned the intentions or professionalism of my colleagues across industry and government. Some of the most intelligent, mission-driven people I know work in this field. They are committed, deeply and sincerely, to the principle of Never Again.

But today, good intentions and institutional rigor are no longer enough.

Over the past several months, I have become intensely focused on artificial intelligence, not because it builds better slide decks or automates workflows (though it does), but because it is redefining the national security landscape in real time. And the pace of change is deeply unsettling.

If you are among those who have said, “I tried AI—it’s not that impressive,” or “It won’t touch our sensitive or classified workflows,” I encourage you to revisit that assumption. Open the most recent versions of leading frontier models released in just the past few weeks. The progress over the last several months alone is staggering. Many experts now anticipate that within a year, these systems will be capable of materially accelerating complex technical design, reverse engineering processes, and operational planning at levels previously reserved for highly trained specialists.

Let that sink in.

For the past quarter century, regulators, technology developers, and operators (airlines, airports, cargo carriers) have focused on identifying hostile intent and threat materials, then building training, deploying new technologies, and implementing countermeasures to reduce risk. Unpredictability has been layered into the system to guard against the feared black swan.

This model has served us well.

But the time-to-deploy cycle between identifying a new threat and implementing mitigation often spans years. Not because we lack engineering talent, but because aviation security quite appropriately relies on rigorous standards-setting, certification protocols, procurement requirements, and oversight mechanisms. Historically, that rigor has been a strength.

Today, it is becoming a vulnerability.

While governments and experts spend months or years developing standards, refining common test methodologies, and navigating procurement cycles, adversaries are accelerating attack-path innovation at unprecedented speed.

Recently, a former FBI colleague demonstrated an AI agent built using open-source code that could assist in designing improvised explosive devices and estimate the likelihood of evading certain detection approaches. This was based solely on publicly available information. Now consider what happens when detection parameters are inferred, modeled, or back-engineered by sophisticated actors using advanced AI systems.

This is not a hypothetical future risk. It is a present reality.

We cannot put the genie back in the bottle. But we can adapt, quickly, deliberately, and collectively.

A Call to Action

For Governments

Privatize Non-Core Detection Evaluation Functions

There are only a handful of true explosives detection evaluation experts within government. Their time should be protected and singularly focused on standards iteration and equipment evaluation. Administrative burdens and peripheral tasks dilute impact. Consolidate expertise. Create dedicated, mission-driven “war rooms” where senior experts are paired with emerging engineering talent and empowered to move quickly, test aggressively, and iterate continuously.

Leverage AI for Accelerated Standards Development

Like-minded governments should establish secure AI-enabled platforms to rapidly develop and refine both achievable and aspirational detection standards. AI can dramatically accelerate consensus modeling, requirements analysis, and trade-off assessment. A small, nimble, data-sharing cohort dedicated exclusively to this effort could compress timelines from years to months or weeks. Collaboration cannot remain a side effort; it must be the mission.

Fund Achievement Through Competitive Sprints

Governments should sponsor recurring, open detection evaluation sprints that incorporate evolving threat streams and synthetic data. Use AI to design scenarios, generate edge cases, and analyze performance. Reward top-performing solutions with rapid funding for limited deployment. Procure in small batches. Field test quickly. Iterate. Replace if necessary. The objective is velocity with accountability.

Reform Incentive Structures

Commodity pricing models that reward only baseline compliance dampen innovation. Requirements should incentivize higher performance. Operators that invest in superior detection capabilities could receive regulatory flexibility, reduced inspection frequency, or relief from layered add-on measures. AI-enabled modeling can help identify incentive structures that drive meaningful performance improvements without undermining oversight.

Prioritize the Human Frontline

As AI strengthens both offense and defense, frontline officers become even more critical. Equip them with real-time training that illustrates how adversaries may leverage AI to evolve tactics. Use AI-generated imagery and scenario modeling to sharpen instinct and situational awareness. Machines will detect patterns. Humans will detect anomalies. Experience tells us that intuition remains irreplaceable.

For Original Equipment Manufacturers (OEMs)

Collaborate Relentlessly

The next generation of threat mitigation will not emerge from isolated development cycles. It will require advanced data pipelines, AI integration, and sensor fusion. Partner with emerging AI-native companies. Combine hardware expertise with computational horsepower. Set aggressive milestones. If governments accelerate standards, industry must be prepared to rise accordingly.

Demonstrate What Is Possible, Before Being Asked

Do not wait for formal standards to prove capability. Use AI and existing hardware platforms to demonstrate performance envelopes and share non-sensitive results publicly. If small, agile teams can prototype advanced capabilities in days using modern AI stacks, established OEMs can, and must, move faster. The market will not wait.

Institutionalize Structured Engagement

Establish recurring, high-level technical sessions between CTOs and government evaluation experts. Replace memos and process critiques with data-driven dialogue. Use AI tools to identify areas of consensus, clarify trade-offs, and accelerate work plans. Accountability must be shared, and progress must be measurable.

For Operators

Own Operational Readiness Testing

Airports and airlines should co-invest in private testing and evaluation environments that assess operational readiness under real-world conditions. Provide physical space and operational context. Think of it as a “Consumer Reports” model for detection technologies that’s objective, data-driven, and transparent within appropriate security boundaries. AI can help analyze performance across diverse operating environments.

Bring Operational Context to Risk Discussions

Governments will continue to manage the classified dimensions of explosives detection. Operators, however, are uniquely positioned to quantify operational risk across the broader system. Collect data. Use AI-driven modeling to demonstrate where collective resources most effectively reduce risk. Engage regulators in substantive, evidence-based discussions aimed at iterative improvement and not static compliance

These recommendations were drafted on a flight between two U.S. cities. I suspect that AI systems could generate even more refined recommendations in the weeks ahead.

But one conclusion is clear: we are entering a period that demands decisive leadership in aviation security and beyond. The pace of technological change is compressing response timelines across every domain of national security.

Speed now matters as much as rigor.

I have unwavering faith in this community, its expertise, its dedication, and its shared mission. The question before us is whether we can match our historic commitment with a new operational tempo.

May we rise to the occasion.