The Deepfake Threat: Defending Truth in the Age of AI

By Shawn Roslin, Vice President of National Intelligence Division and International Programs

In this new landscape, national security is not just about military strength or economic power. It is about the ability to defend truth, maintain trust, and operate effectively in an environment where seeing is no longer believing.

The World Economic Forum’s 2025 Global Risks Report ranks misinformation and disinformation among the top global risks, highlighting the increasingly blurred lines between AI-generated and human-created content.

National Security Implications

From a national security perspective, deepfakes present a low-cost, high-impact weapon. Adversaries can deploy them rapidly, anonymously, and at a large scale.

The Department of Homeland Security’s report, "Increasing Threat of Deepfake Identities", notes that fake media is often used to manipulate trust, with a significant percentage of deepfakes tied to malicious or non-consensual use.

Consider the potential scenarios:

• A fabricated video of a U.S. official announcing military action
• A falsified audio clip triggering financial market panic
• Deepfake communications used in cyber or espionage operations

These are not hypothetical risks—they are emerging realities.

In January 2024, a Hong Kong finance employee joined a video call with individuals they believed to be the company’s UK-based CFO and colleagues, only to be deceived by real-time AI-generated deepfakes and transfer $25.6 million.

The growing concern over deepfakes has prompted governments to treat artificial media as a national security issue instead of a technological challenge. In February 2026, the United Kingdom announced a collaboration with Microsoft, law enforcement agencies, academics, and international partners to develop a world-first deepfake detection evaluation framework. The initiative was launched in response to the rapid growth of AI-generated content, with an estimated 8 million deepfakes shared in 2025 alone, up from 500,000 in 2023. The framework will test detection technologies against real-world threats, including election interference, organized crime, fraudulent documentation, impersonation, and other risks to public safety and national security.

As geopolitical competition intensifies and nations race to harness emerging technologies, AI-driven disinformation and deepfake-enabled influence operations are increasingly likely to become standard tools of statecraft.

The Speed Problem: Technology vs. Defense

One of the most pressing challenges is the pace of advancement. Deepfake generation tools are becoming more accessible, more sophisticated, and harder to detect.

Meanwhile, detection technologies lag behind. Even the most advanced systems struggle to keep up with rapidly improving generative models. This creates a dangerous window of vulnerability, leaving room for malicious actors to exploit the technology faster than defenses can adapt.

Policy Gaps, Preparedness, and Public Awareness

Addressing the deepfake threat requires a multi-layered response:

• Technological investment in detection and authentication systems
• Government policy frameworks to regulate misuse without stifling innovation
• Public education to improve media literacy and resilience

Recent developments highlight the growing tension between artificial intelligence companies and government agencies over how this technology should be used. In a widely reported case, AI firm Anthropic publicly refused to allow its technology to be used by the Pentagon for mass domestic surveillance or fully autonomous weapons systems. The dispute underscores a broader challenge: while governments seek to leverage AI for national security and intelligence purposes, the legal and ethical frameworks governing its use remain underdeveloped. Current laws permit large-scale data collection on U.S. citizens, but the integration of AI could dramatically expand the speed, scale, and precision of surveillance—raising serious concerns about civil liberties, oversight, and misuse. This same gap between technological capability and regulatory oversight is what enables the rapid spread and weaponization of deepfake content.

Current legislation addressing deepfakes focuses on prohibiting non-consensual intimate imagery, curbing electoral disinformation, and protecting national security from AI-manipulated content.

This year, the TAKE IT DOWN Act was signed into law, targeting non-consensual intimate imagery, including AI-generated deepfakes. Additional proposals include:

• The NO FAKES Act, aimed at protecting individuals’ voice and likeness from unauthorized AI-generated replicas
• The DEEPFAKES Accountability Act
• The Deepfake Report Act, focused on identifying foreign threats

A few defense and technology security measures in place are:

• Deepfake Report Act of 2019/2021: Requires the Secretary of Homeland Security to publish annual reports on how deepfakes are used to undermine national security and elections.

• Defense & Intelligence Focus: Emphasizes investment in detection technologies to counter foreign disinformation
• AI-Generated Evidence in Court: Proposals to strengthen authentication standards including the Federal Rule of Evidence 901

Equally important is coordination between public and private sectors. Social media platforms, intelligence agencies, and news organizations must work together to identify and respond to threats in real time.

A Fight for Truth Moving Forward

The rise of deepfake technology marks a turning point in the information age. The question is no longer whether AI can manipulate reality—but how societies will respond when reality itself becomes contested.

Failure to address this challenge risks ceding a critical domain of modern conflict—one where perception shapes reality, and reality itself can be engineered.