"I think right now we’re looking at how to use an AI where we could triage so much more," David Imbordino said on a panel at the Billington Cybersecurity Summit in Washington, D.C.
David Imbordino on AI triage
Imbordino, head of the National Security Agency’s cybersecurity directorate, told officials from the “Five Eyes” alliance that the agency is leaning into artificial intelligence to process the volumes of data it ingests every day. He framed the goal in practical terms: move analysts from manual slog to rapid sorting so they can find unusual events and prioritize follow-up. As Imbordino put it, “volume has always been a problem,” and AI raises the prospect of getting “to the nuggets” faster than before.
NSA cybersecurity directorate: collection and defense
The directorate Imbordino leads focuses on both intelligence collection and cyberdefense. The agency routinely ingests large datasets, the source says, including intercepted foreign communications and electronic signals from radar and weapons systems. Those disparate streams are the very information AI tools are being asked to join up and evaluate more quickly than current methods, which Imbordino said can take days or weeks.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildSpeed: from days or weeks to "much faster"
Officials described a concrete change in workflow: tasks that today can consume days or weeks would be done, in Imbordino’s words, “much faster.” The emphasis was not on replacing analysts but on augmenting them — letting algorithms triage and surface anomalies so human experts can focus on interpretation, context and decision-making. The claim in public remarks is straightforward: AI can collapse long analytic timelines and expose higher-value signals buried in mass data.
Tim Kosiba and the agency’s pursuit of model access
Late last month, NSA deputy director Tim Kosiba said the agency wants access to a wide range of models and is engaging heavily with frontier AI companies. That stance dovetails with the directorate’s operational needs: access to diverse models could expand the range of analytic techniques available for signal correlation, anomaly detection and vulnerability discovery, according to the agency’s public comments.
The June executive order: classified tests and a 30‑day window
The agency is also helping implement a June executive order directing national security agencies to create classified tests for determining when AI systems have developed sufficiently advanced hacking capabilities to warrant additional scrutiny. The order calls, in part, for a voluntary arrangement through which developers can give the federal government access to qualifying models for as many as 30 days before releasing them to other trusted partners. NSA officials described both the classified testing regime and the voluntary pre-release access as central to how national security agencies will try to assess and manage rapidly advancing tooling.
How frontier AI companies, national security agencies, and defenders are responding
- Frontier AI companies: the agency is engaging heavily with these firms and the executive order contemplates a voluntary mechanism for developers to provide models to the government for up to 30 days before broader release.
- National security agencies: the NSA is helping implement the June order’s classified tests to measure when AI systems reach thresholds that require extra scrutiny, and agency leaders are pressing for access to a wide range of models to support analytic and defensive missions.
- Defenders and analysts: Imbordino said AI tools will be used to triage information and evaluate disparate pieces of data “much more quickly,” turning processes that today take “days or weeks” into workflows that are “much faster.”
The picture painted in these remarks is immediate and operational: an agency accustomed to coping with vast, varied signals sees AI as a force multiplier for triage and analysis, and it is pursuing both technical access to models and policy mechanisms to assess their risks. NSA leaders are moving on two tracks at once — buying or developing capability and working within the framework of a June executive order that mandates classified testing and contemplates a 30‑day access window — leaving the concrete next steps to implementation and to whatever agreements emerge between the government and frontier AI companies.
