The primary threat from current AI is not hallucination but intentional curation. Models designed to hide specific topics are fundamentally untrustworthy because they actively lie by omission. By selectively narrowing the universe of information, the AI becomes a subtle, constant manipulator.

The ability to label a deepfake as 'fake' doesn't solve the problem. The greater danger is 'frequency bias,' where repeated exposure to a false message forms a strong mental association, making the idea stick even when it's consciously rejected as untrue.

Political strategist Bradley Tusk warns that the tech industry is in a bubble regarding public perception of AI. He predicts AI will be a major target in upcoming elections, blamed for both job losses and rising energy prices from data centers. Challengers will use anti-AI sentiment as a powerful tool against incumbents, a reality most in tech are not prepared for.

The modern information landscape is saturated with AI-generated propaganda from all sides. It is no longer sufficient to be skeptical of foreign adversaries; one must actively question and verify information from domestic governments as well, as all parties use these tools to shape narratives.

A content moderation failure revealed a sophisticated misuse tactic: campaigns used factually correct but emotionally charged information (e.g., school shooting statistics) not to misinform, but to intentionally polarize audiences and incite conflict. This challenges traditional definitions of harmful content.

The biggest political danger of deepfakes isn't that people will believe fake content. It's the "liar's dividend": politicians can now dismiss genuine, scandalous video evidence as a deepfake. This erodes video as a tool for accountability, a more subtle but profound threat to political discourse.

Because the general public poorly understands AI, the topic becomes a blank canvas for political manipulation. Politicians can create any perception they want—from job-stealing menace to national security threat—to shape opinion and move votes.

While deepfakes garner attention, research from as early as 2020 shows AI can measurably change political opinions using only simple text. This scalable, text-based persuasion is a potent tool for information operations that may be more impactful than more technologically complex manipulations.

Beyond generating fake content, AI exacerbates public skepticism towards all information, even from established sources. This erodes the common factual basis on which society operates, making it harder for democracies to function as people can't even agree on the basic building blocks of information.