The first article in this series described how campaigns are using AI right now. This one covers who is supposed to govern it, and how far apart the answer-givers are.
The federal picture shifted early in 2025. President Trump signed executive orders revoking Biden-era AI restrictions and establishing a new federal framework, according to the White House. The administration framed the move as clearing obstacles to American innovation. Critics argued it removed accountability structures before replacements were in place.
What followed was a direct collision between federal ambition and state action.
When the Senate considered language that would have blocked states from enforcing their own AI laws for ten years, it failed 99-1, according to Biometric Update. That outcome sent a clear signal: Congress isn't ready to preempt state authority here. States had already moved aggressively on their own. The National Conference of State Legislatures tracked more than 1,000 AI-related bills introduced across state legislatures in 2025 alone. At least 30 states now have laws specifically targeting AI-generated content in elections, according to GovFacts.
Not all of those laws survived legal review. California's AB 2839, which targeted AI deepfakes in political advertising, was partially struck down as unconstitutional, according to analysis from Skadden. The line between election protection and free speech remains unsettled in court.
Public opinion, meanwhile, is ahead of both branches.
A survey from Elon University found that 78 percent of Americans expect AI to be used to abuse the political process in some way. An AP-NORC poll for PBS found that by an 8-to-1 margin, voters believe AI will harm elections rather than help them. Support for regulation crosses party lines: a Public Citizen survey found 76 percent of voters want AI regulated in politics, including 71 percent of Republicans.
The complication is trust, not agreement. Pew Research found that 54 percent of Republicans trust the United States government to regulate AI responsibly, compared to 36 percent of Democrats. Both figures are low. They reflect a shared skepticism about whether any institution is currently equipped to move fast enough to matter.
Journalism sits in a difficult position inside all of this.
The Reuters Institute found that only 12 percent of readers are comfortable with news produced entirely by AI without human involvement. But the economics of political media are pushing outlets toward automation regardless. A study published on arXiv found that nearly 90 percent of journalists believe AI will significantly increase the risk of disinformation in political coverage. The reporters covering campaigns are not optimistic about what the technology does to the information voters rely on.
There is no clean resolution here. The technology is outrunning the regulations. The regulations are outrunning enforcement. And enforcement is outrunning public understanding.
The decisions being made right now, in courts, in statehouses, and inside campaign war rooms, will define what political competition looks like for years ahead. The legal and regulatory frameworks being built, or not built, in the next few years will either create accountability or create a vacuum. Voters who want a role in that outcome will need to pay attention.
This is Part 2 of a two-part series. Read Part 1: AI Has Already Changed How Campaigns Win.