Artificial Intelligence vs. Truth

Photo by Igor Omilaev on Unsplash

One of the most destructive agents of disinformation in our age is artificial intelligence (AI), which can create and disseminate untruths around the world almost instantaneously. It’s becoming increasingly difficult to separate fact from “manufactured truth.” AI’s generative tools are capable of producing text, images, audio, and video which profoundly alter how people encounter information. AI does offer real benefits—for research purposes, in particular. However, its prolific use in producing “original” content (from existing content it harvests online) has eroded public trust in what we see and read. Sowing distrust is a key tool of authoritarian regimes. “Don’t believe your eyes or ears,” autocrats say. “Believe what we tell you.”

The line between authentic and synthetic content is increasingly blurred, creating what researchers call a “liar’s dividend”: when anything can be faked, everything becomes easier to doubt. Understanding why and how this erosion of trust is happening is essential to protecting the integrity of information ecosystems and democratic discourse.

One of the primary drivers of declining trust is the rise of highly convincing synthetic media. AI technologies have lowered the barrier to creating deceptive content. What once required advanced technical expertise can now be done quickly and cheaply, often by individuals with minimal training. This shift has had a direct impact on how people perceive visual evidence. Historically, photographs and videos were treated as strong indicators of truth. While manipulation has always existed, it was relatively difficult to execute convincingly. Today, AI-generated images can depict events that never occurred—political figures in fabricated scenarios, disaster scenes that never happened, or “evidence” of crimes that were never committed. As these images circulate widely on social media, they contribute to confusion and skepticism. Even when authentic visuals are presented, audiences may question their legitimacy.

Deepfakes

The phenomenon of deepfakes illustrates this challenge clearly. Deepfakes are AI-generated or AI-altered videos that make individuals appear to say or do things they never did. Research from organizations such as MIT Media Lab has shown that humans struggle to reliably distinguish deepfakes from real videos, especially as the technology improves. A 2022 study by the Stanford Internet Observatory highlighted how quickly synthetic media tools are evolving, warning that detection methods often lag behind generation capabilities.

This technological asymmetry—where creating fakes is easier than detecting them—undermines trust. When viewers know that convincing fakes exist, they may begin to doubt all visual content, not just suspicious examples. This is the essence of the “liar’s dividend.” Public figures, for instance, can dismiss genuine evidence as fake, exploiting general uncertainty to avoid accountability. In this way, the existence of AI-generated deception weakens the evidentiary power of real media.

Text-based AI has contributed to the erosion of trust as well. Large language models can produce fluent, persuasive writing that mimics human authorship. While useful in many contexts, these systems can also generate misleading articles, fabricated citations, or coordinated disinformation at scale. AI-generated content can amplify existing misinformation campaigns, making them more difficult to detect and counter.

Another factor contributing to declining trust is the speed and scale at which AI-generated content spreads. Social media platforms prioritize engagement, often amplifying content that is novel, emotional, or controversial, qualities that synthetic media can easily replicate. Once misleading content goes viral, corrections often fail to reach the same audience. Studies by the Pew Research Center have shown that Americans are increasingly concerned about their ability to identify accurate information online, with many expressing uncertainty about what sources to trust.

Consequences of trust erosion are far-reaching. In journalism, audiences may become skeptical of legitimate reporting, weakening the role of the press as a watchdog. In politics, voters may struggle to distinguish real statements from fabricated ones, undermining informed decision-making. In everyday life, individuals may become cynical, adopting a default assumption that all information is suspect. This uncertainty can be just as damaging as believing falsehoods outright, as it discourages engagement and erodes shared reality.

AI Regulation

Efforts to regulate Artificial Intelligence have been spotty and inconsistent. Most states limit use of AI in certain specific situations, such as employment decisions, or in certain industry sectors, such as healthcare. Many states are updating existing privacy laws to include rights for individuals to access, correct, or delete data used in AI profiling. As AI technologies develop, there will undoubtedly be new and more specific laws passed in response to disturbing or outrageous situations. While it might be hoped that the industry would police itself, that cannot be assumed. Remember Grok’s explicit photo generation and Musk’s unhelpful response?

Texas enacted the Texas Responsible Artificial Intelligence Governance Act (TRAIGA) via House Bill 149 on June 22, 2025, to regulate AI use. It was effective January 1, 2026 and controls AI development and deployment by prohibiting harmful uses, such as intentional discrimination, behavioral manipulation, and unlawful deepfakes. It establishes a regulatory sandbox for testing AI, requires government transparency, and bans non-consensual biometric surveillance The law also establishes the Artificial Intelligence Advisory Council to study and monitor use of AI systems in state government. Stay tuned, though. Greater oversight will almost surely be warranted.

The Trump administration’s emphasis is on innovation in AI, rather than regulation. President Trump’s December 11, 2025, Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence,” aims to accelerate AI development by reducing regulatory barriers and establishing a uniform federal standard over a “patchwork” of state AI laws. The order emphasizes AI dominance, promotes industry-led standards, and threats to withhold federal funding from states with restrictive AI regulations. There are efforts to accelerate innovation while implementing safety standards, however. Key initiatives include the White House National Legislative Policy Framework (2026), focusing on child safety (Kids Online Safety Act), intellectual property protection (NO FAKES Act), and restricting state-level AI regulations to create a single national standard. It remains to be seen whose interests will be protected in such a standard: AI developers, AI investors, or ordinary citizens.

How to Detect Inauthentic Information

Despite challenges posed by eroded trust in information sources, there are practical steps individuals can take to better distinguish AI-generated content from authentic material. While no method is foolproof, a combination of critical thinking and verification techniques can significantly reduce vulnerability to deception.

1. Examine the source. Content from established, reputable organizations with transparent editorial standards is more reliable than anonymous or unfamiliar accounts. Cross-checking information across multiple credible sources can help confirm its accuracy.

2. Look for inconsistencies in visual details. AI-generated images often contain subtle anomalies—unnatural lighting, distorted text, irregular reflections, or mismatched shadows. In videos, pay attention to unnatural facial movements, inconsistent lip-syncing, or visual artifacts around the edges of a subject’s face.

3. Verify context using reverse image search tools. Google Images or TinEye can reveal whether an image has appeared elsewhere online, sometimes in a different context, helping identify reused or manipulated content.

4. Consider the plausibility of the content. Ask whether the scenario depicted aligns with known facts and credible reporting. AI-generated disinformation often relies on emotionally charged or sensational claims designed to provoke immediate reactions.

5. Rely on emerging verification technologies. The Content Authenticity Initiative aims to embed metadata in digital media, providing information about how and when content was created. While not yet universal, it represents a step toward restoring trust.

6. Cultivate a habit of deliberate skepticism rather than reflexive disbelief. The goal is not to assume everything is fake, but to approach information thoughtfully, recognizing both the possibilities and limitations of AI-generated content.

The proliferation of AI has complicated the information landscape. By making it easier to fabricate convincing text, images, and videos, it has weakened long-standing assumptions about the reliability of sensory evidence. The solution is not to abandon trust altogether, but to rebuild it on a more informed foundation. Through education, critical evaluation, and technological safeguards, individuals and institutions can adapt to this new reality.

Trust, once eroded, is difficult to restore. But with awareness and deliberate effort, it is possible to navigate an AI-saturated world without losing confidence in the existence of truth itself.