For decades, Democrats have held to Michelle Obama’s principle: “When they go low, we go high.” The high road has been more than a slogan — it’s been a strategy, a way to preserve integrity in the face of mudslinging. But in the era of generative AI, the high road is becoming an impossible path to walk.

The Fragility of the High Road

The high-road strategy depends on credibility. A candidate must project dignity and restraint, showing voters they are above petty attacks. But generative AI has introduced an asymmetry. With the ability to manufacture convincing deepfakes of candidates saying foolish or offensive things, bad actors can weaponize perception itself. One viral clip, even if fabricated, can erode the trust that the high road requires.

Worse, the asymmetry is one-sided. A candidate committed to the high road risks alienating their own base if they descend into disinformation. Their opponent, however, suffers no such penalty if their base is comfortable with “low road” tactics. Add to this the amplification power of social media algorithms — which reward outrage, not nuance — and the result is a battlefield tilted against those who value integrity.

Generative AI as Election Interference

This isn’t just politics-as-usual. Generative AI represents a new form of election interference. Whether deployed by foreign adversaries or domestic partisans, deepfakes destabilize the very foundation of democracy: a shared baseline of truth.

Deepfakes are not persuasion. They are deception, deliberately designed to bypass reason and exploit emotion. That makes them more akin to voter suppression or ballot fraud than to campaign rhetoric. Framing AI fakes as election interference is essential to helping the public understand the gravity of the threat.

Lessons From Cigarettes: Warning Labels for Democracy

We’ve faced industries before that profited by deceiving the public. Cigarette manufacturers once downplayed health risks while addiction spread. The solution was not to ban tobacco outright, but to mandate clear, unavoidable warnings on every package and advertisement: “Smoking Kills.”

We need the same approach for generative AI in politics. Social media platforms should be required to put large, standardized warning labels on AI-generated content involving political figures. Not fine print, not subtle disclaimers — bold, screen-dominating notices that alert viewers:

WARNING: This content is AI-generated and may misrepresent reality.

These warnings should appear before playback, remain visible as watermarks, and be enforced across platforms. Just as cigarette labels shifted public norms and stigmatized smoking, democracy warnings on AI content can stigmatize manipulation and help voters build resistance.

Building the Secure Road

If the high road is to survive, it must evolve. Integrity alone won’t withstand an onslaught of synthetic lies. The path forward is a “secure road” strategy:

• Pre-bunking: Educating voters ahead of time about deepfakes.
• Rapid response: Fact-checking fakes within hours, not days.
• Reframing: Exposing the use of deception as a sign of weakness, not strength.
• Standardized warnings: Making AI manipulation visible and stigmatized, like cigarettes.

A Call to Action

No matter your political stripe, generative AI deepfakes threaten your right to make an informed choice. Left unchecked, they will erode trust not only in candidates but in democracy itself.

We cannot leave this fight to individual campaigns. Social media platforms, regulators, and legislators must act now to ensure that synthetic political content is clearly and unmistakably labeled. Without such protections, the high road will collapse under the weight of deception — and voters will be left wandering a fog of lies.

Because lies, like smoke, are toxic. And our democracy has a right to breathe clean air.