"We can’t compromise cybersecurity for speed. That will only impact us in a very negative way, especially if [AI] is tied to critical infrastructure," Michael Montgomery, president of Chenega Architecture and Design Solutions, warned on the Government Technology Insider podcast.
Michael Montgomery’s warning on cybersecurity and speed
The podcast frames Montgomery’s warning as central to how federal agencies should approach AI. His statement — quoted directly on the show — ties a clear constraint (cybersecurity cannot be sacrificed for deployment speed) to a concrete risk vector (AI systems connected to critical infrastructure). That coupling grounds the episode’s broader argument: decisions made early in acquisition and implementation can create systemic vulnerabilities later.
Power, cooling, and the physical infrastructure for AI
The series’ first episode focused on a foundational point: the physical infrastructure required to power and cool systems that enable AI use cases is a priority for agency leaders. Montgomery, in that initial installment, explained how agencies can address powering and cooling to support mission-enabling AI capabilities. The current episode builds from that technical foundation and turns attention to operational concerns that sit atop it.
Acquisition, implementation and design are linked
The podcast emphasizes that the way solutions are acquired, implemented, and used has a direct impact on the design of AI systems. In other words, procurement choices and operational practices are not peripheral; they materially shape the systems’ technical architecture and the protections those systems will need. Trust, security, and procurement are presented as interconnected elements that determine the value agencies ultimately get from AI — and so, per the episode, they must be considered from the outset of any program.
Treating AI as an operational entity: continuous monitoring and shutdown triggers
Montgomery framed AI not as a one-off install but as an ongoing operational entity. “AI has to be treated as an operational entity that requires continuous monitoring; it’s got to have safeguards and defined triggers in place to shut it down should it start to act in a way that is malicious or an issue to the system,” he said on the podcast. That language places emphasis on lifecycle controls: detection, automated or human-in-the-loop responses, and pre-defined shutdown conditions. The episode positions these measures as necessary complements to the physical infrastructure work covered in the prior installment.
What this means for technologists, procurement leaders, and security teams
- Technologists and security teams: Montgomery’s formulation implies they should plan for ongoing operational duties — continuous monitoring and safeguards — rather than treating AI deployments as static projects. The podcast makes clear these teams will need to integrate monitoring and shutdown mechanisms from day one.
- Procurement leaders: Because the way systems are acquired and implemented affects system design, procurement decisions are described as strategic design decisions. The episode underscores that procurement processes must account for trust and security requirements as part of acquiring mission-enabling AI.
- Agency leaders responsible for infrastructure: The earlier episode’s focus on power and cooling is reiterated here as a reminder that operational resilience begins with the physical layer; investments in powering and cooling are foundational to secure, usable AI.
The Government Technology Insider podcast connects two practical threads: the physical systems that enable AI, and the operational controls that keep those systems trustworthy and secure. Montgomery’s central admonitions — do not trade cybersecurity for speed, and treat AI as an operational entity with continuous monitoring and shutdown triggers — frame the episode’s advice for federal deployments. For listeners and decision-makers seeking more detail, the show offers a deeper discussion of the operational challenges agencies must keep in mind as they invest in AI.
Listen to the full podcast here: https://governmenttechnologyinsider.com/trust-security-and-procurement-deploying-ai-the-federal-way/
