"I have been plagued by my own staff hijinks on the use of artificial intelligence," Adm. Frank Bradley told the Global SOF Foundation’s Indo‑Pacific Irregular Warfare Symposium in Honolulu, using a small, wry anecdote to signal a much larger problem: AI is changing how information is made, analyzed, and weaponized, and that change is eroding trust across military and civilian systems.
Adm. Frank Bradley on AI, deception, and unity of effort
Bradley, the leader of U.S. Special Operations Command, argued that artificial intelligence is a double‑edged sword. On one hand, faster intelligence gathering and analysis is "allowing special operators to pull off missions that would have been impossible years ago." On the other, "more persuasive AI‑enabled content, quietly but consistently shaping perceptions about what is happening where, has become the dominant feature of a new critical battleground."
He warned that the result is a loss of confidence in digital information that will not only complicate individual operations but also "affect...the ability of the United States to 'generate that unity of effort' to deter or defend against adversarial attack." Bradley framed early, credible exposure of hostile behavior as a core advantage—“Information advantage leads to decision advantage.”
Faster intelligence synthesis and the April rescue of a downed F‑15 pilot
Bradley pointed to a concrete example of AI‑enabled gains: the April rescue of a downed F‑15 pilot behind enemy lines. He described that operation as "made possible by a tightly synchronized intelligence, cyber‑enabled support, and disciplined coordination across multiple elements." That kind of mission, he said, depends on a common intelligence picture that updates as quickly as real‑world conditions.
An intelligence official at the event described how the tools have changed who sees what and how quickly. About a decade ago, advanced virtual imagery and fused data streams were limited to specialists; now, Bradley and the official said, teams bring "10, 20, 30—I mean, dozens, if not thousands, of different data sources in to build that operator level," with some feeds updating "every 15 minutes."
Audit-ready is a season. It shouldn't be.
Evidence in spreadsheets, controls drifting between audits, frameworks multiplying on flat headcount. Nubivance runs continuous compliance on Rapid7 Cyber GRC - SOC 2, HIPAA, ISO 27001, PCI, CMMC.
End the scrambleActive measures amplified: Russia, China, and Ibrahim Traoré
The symposium speakers tied the dangerous side of AI to long‑standing adversary tactics. Ongoing Russian and Chinese efforts to deceive Western audiences—often referred to as "active measures"—are being amplified by readily available AI tools. A former State Department intelligence official put it bluntly: "Active measures have eaten the world."
That official also pointed to Ibrahim Traoré, the leader of a junta controlling Burkina Faso, as an example of how even smaller actors can employ AI tools to shape domestic populations. The point was not scale alone but accessibility: simple AI tools can produce deepfakes and synthetic content that make deception far easier for a wide range of actors.
Erosion of U.S. detection tools and social‑platform transparency
The former State Department official argued that detection and deterrence capacities inside the U.S. government have been weakened "under the new Trump administration," saying the United States "had surrendered critical tools to watch and detect foreign influence just as adversaries were ramping up their ability to conduct influence operations." Changes in social media company leadership, the official said, have decreased transparency into content moderation and allowed more disinformation on platforms.
That official specifically referenced the loss of structures that once protected the domestic environment: the Foreign Influence Task Force (FITF) at the FBI, the Foreign Malign Influence Center at the DNI, and "various USAID activities and programs." The consequence, they warned, is practical and cultural—analysts outside those entities may be less likely to raise concerns, because "the federal bureaucracy is very sensitive to priorities of political leadership" and examining influence operations "is not going to be a career‑enhancing move." In their words, "We're in this period now where we are largely flying blind."
What this means for technologists, policymakers, and special operators
- Technologists and security teams: Bradley described active work "inside the the Department of War, but across the United States government as well, to ensure that those data sets and those those systems are are appropriately protected against agentic AI attacks." Teams will need to treat datasets and fusion systems as primary targets for adversarial manipulation.
- Policymakers and platform overseers: The former State Department official highlighted how changes at social platforms and within government task forces have reduced transparency and detection. Restoring or replacing those lines of sight—whether inside agencies or through platform accountability—was presented as a necessary policy horizon.
- Special operators and commanders: The promise of near‑real‑time, fused intelligence was clear in the F‑15 rescue example, but so was the risk. A U.S. military official working in counter‑intelligence and foreign influence analysis summed the dilemma: "AI has made deception scalable... we're no longer just fighting phishing emails, we're fighting synthetic voices, cloned faces, and entire fake identities convincing enough to fool the people who know the real person best."
Conclusion: credibility decay as an operational threat
Speakers at the symposium sketched a stark tradeoff: AI can compress the sensor‑to‑decision loop and enable rescues and synchronized operations in ways previously impossible, yet the same technologies scale deception and undercut the trust that makes joint action possible. Bradley said the government is "spending a good bit of time and energy" to harden systems against "agentic AI attacks." Whether those protections can keep pace with rapidly improving synthetic content, platform opacity, and a shifting organizational posture inside government remains the central unresolved question the participants left on the table.
