A federal judge has struck down California’s first-in-the-nation effort to regulate AI in elections, setting off a clash between free speech and election integrity just as the 2026 cycle approaches.

The ruling swept aside deepfake restrictions and platform takedown mandates, placing California’s attempt to police AI-generated political content on a direct collision course with the First Amendment.

Governor Gavin Newsom and lawmakers had warned that generative AI could make democracy more “fraught,” flooding voters with fake videos, images, and robocalls designed to manipulate behavior.

Judge John Mendez did not dispute those risks. Instead, he ruled that the state cannot respond by criminalizing political speech or compelling platforms to remove content.

By striking down both the content-based ban and the takedown requirement, Mendez leaned on longstanding free-speech doctrine and Section 230, which limits state authority over online platforms.

As a result, free speech advocates are celebrating, while election-integrity supporters are alarmed. Unlikely players like The Babylon Bee, X, and Rumble emerged as symbols of constitutional protection for even deceptive expression.

The decision leaves California entering its first generative-AI election cycle with few tailored safeguards, despite lawmakers’ fears of widespread voter confusion and manipulation.

Now the burden shifts to voters, campaigns, and platforms to develop norms of verification and skepticism, forcing democracy to confront AI-driven chaos in the open—and endure it.