Colorado could be the next state to set guardrails on how artificial intelligence is used in political campaigns and election communications, as the world adapts to new technologies that can produce realistic videos, audio, and images far removed from reality.
Colorado’s top election official said the goal is to try to better inform the public about what’s happening so they can figure out what’s real and what’s not.
“AI is being used to manipulate what candidates or elected officials say or do to make them appear like they're doing something or saying something that they actually did not,” said Democratic Secretary of State Jena Griswold.
The legislation, which is still being finalized before it’s officially introduced, would require a disclaimer on any artificial intelligence-generated electoral content that’s distributed in Colorado.
“AI technology is so sophisticated that sometimes it's really hard to discern whether a communication is fake or not,” said Griswold.
While the oft-foretold boom in AI-driven election chaos has been slow to materialize, there has been at least one potential case lately. New Hampshire’s Attorney General is investigating robocalls that appeared to be artificially created to mimic President Joe Biden, urging people not to participate in that state’s primary.
Democratic Rep. Brianna Titone of Arvada will be one of the Colorado bill’s main sponsors. She said details on the disclosure requirement are still being worked out, but it would apply to all types of communication and would need to be an audio disclosure for robocalls, and television and digital ads, not just written.
Titone said AI content is especially challenging to track in the social media sphere where it’s hard to figure out the origins of a post.
“And even if that was created with a disclosure, (others) may take away that disclosure (when re-sharing). They may cover it up or remove it through different digital means to make it look like it was actually real. So there's a lot of things that voters need to be really careful about this election season,” said Titone.
Failure to disclose that content was created using artificial intelligence could incur a financial penalty.
Other states have made efforts in recent years to regulate AI election content, with mixed results. Michigan, Minnesota, and Washington state all passed legislation last year to address the issue. But proposals to require disclosure have failed elsewhere, including Indiana, Illinois, and New York according to the National Conference of State Legislatures.
“Those are the kinds of things that, when done in proximity to an election, can upend what people want to do in that election, how they vote if they vote, and that is a threat to democracy,” said Titone. She’s not just worried about local actors, but that foreign adversaries might take advantage of this technology to create disinformation and try to influence U.S. elections.
However, Titone noted that legislation can only do so much to limit how much-manipulated content is created and distributed. And because any restrictions could potentially butt up against the First Amendment, lawmakers will have to be careful with how the bill is crafted.
Separately, Secretary Griswold is also pushing for a new law to add penalties for anyone who acts as a fake elector in future presidential elections. Some states have prosecuted individuals who submitted certificates naming themselves to the Electoral College based on false claims that former President Donald Trump won the 2020 election.
“We want to make sure that it's very clear that if someone tries to do this, they will be facing major penalties,” said Griswold.
Democratic Rep. Lorena Garcia of Adams County said she decided to sponsor the false electors bill because she wants to protect the fundamental safeguards of democracy.
“We know that Trump's camp is going to go all out in the 2024 election, and we just want to make sure we set our state up for continued success in safe elections,” she said.
