Connecticut lawmakers are trying to address a real and growing problem. Advances in artificial intelligence have made it easier than ever to create convincing “deepfake” videos capable of misleading voters, particularly in the final weeks of an election. That threat deserves serious attention.
HB 5342, advanced this month by the Government Administration and Elections Committee, attempts to respond by restricting the distribution of certain manipulated images, audio, or video within 90 days of an election or primary — specifically when that content is intended to influence the outcome.
The goal is clear: limit deceptive media at the most sensitive point in the election cycle.
But the way the bill attempts to do that raises a different set of concerns.
Rather than narrowly targeting clear cases of fraud or impersonation, HB 5342 applies broad and subjective standards to political speech itself, leaving key questions about legality to be sorted out after the fact.
A Broad Definition of Political Speech
The bill is framed as a response to deepfakes, but its scope extends far beyond that.
Those concerns were raised during the March 20 GAE Committee meeting.
The justification for expanding these restrictions is straightforward: existing legal tools can be difficult to use in fast-moving election contexts. But difficulty alone isn’t a sound basis for expanding government power over political speech.
The bill applies to any content distributed with the intent to “injure a candidate or influence the result of such election or primary.” That language is not limited to deception; it captures the core purpose of political speech itself: persuasion.
The definition of “deceptive synthetic media” is similarly expansive. It includes any image, audio, or video that a “reasonable person” would believe depicts something that did not actually occur.
That standard raises immediate questions. As lawmakers noted during the committee hearing, people vary widely in their ability to identify manipulated content. What appears obviously altered to one viewer may seem convincing to another.
Rep. Seth Bronko (R-Naugatuck) highlighted this concern directly, questioning how the “reasonable person” standard would be applied in practice. The bill offers little clarity.
The Problem of Intent and Reach
The bill also hinges on intent, specifically, whether someone intended to distribute content to an audience of more than 1,000 people.
But in today’s digital environment, reach is often determined not by intent but by algorithms.
During the hearing, Sen. Rob Sampson (R-Wolcott) raised a practical scenario: you post satire to a limited audience. It goes viral, reaching thousands, because algorithms, not intent, determine reach. Under the bill, that could expose individuals to liability based on outcomes they did not anticipate.
When pressed on where the line is drawn, the answer was largely that courts would determine it after the fact.
That creates uncertainty, not just for bad actors, but for anyone engaging in political speech.
Expansive Enforcement Powers
The bill’s enforcement structure adds another layer of concern.
HB 5342 does not stop at civil liability. It introduces potential criminal penalties, including misdemeanor and felony charges for repeat violations. It also allows candidates and private individuals to seek damages, injunctions, and attorney’s fees.
Additionally, the bill authorizes the Attorney General to seek court orders to block the distribution of content.
That kind of authority is typically reserved for clear cases of fraud or consumer scams. Here, it would place the state in the position of determining, in real time, whether certain political communications can continue to circulate.
Combined with the bill’s broad definitions, that raises concerns about discretionary enforcement.
Concerns Across Stakeholders
These issues are not limited to a single perspective.
The Connecticut Business and Industry Association (CBIA) warned that the bill “creates substantial legal uncertainty, compliance burdens, and litigation exposure for businesses operating in good faith.”
It added that the standard of “should reasonably know” introduces a negligence-based framework into a politically sensitive environment where content is created, edited, reposted, and amplified rapidly.
These aren’t hypothetical concerns; they have real consequences. When the rules are vague and the risk is legal trouble, local media, small publishers, and even businesses won’t take the chance. They’ll just stay out of political coverage altogether.
During the hearing, lawmakers raised the same issues. Sen. Sampson warned that “there’s provisions in our law that effectively prohibit us from making laws that are so vague that they have to be so subjective all the time.”
Rep. Bronko questioned how liability works when people reasonably, but incorrectly, believe content is real.
They didn’t get clear answers.
The American Civil Liberties Union of Connecticut (ACLU) raised similar concerns, noting that the bill’s “broad language and vague provisions risk imposing undue restrictions on constitutionally protected speech, particularly political discourse.”
They cautioned that the law “opens the door for selective enforcement, disproportionately targeting political speech that the government may find objectionable.”
A More Targeted Approach
There is a legitimate role for government in addressing election-related deception. But effective policy requires precision.
Existing laws already address defamation, fraud, and intentional interference with elections. If gaps remain, they can be addressed directly, for example, by targeting impersonation of candidates or election officials, or false information about voting procedures.
Those are clear, enforceable lines.
HB 5342 takes a different approach. Rather than focusing narrowly on specific harms, it casts a wide net over political speech, relying on subjective standards and post hoc enforcement.
The Tradeoff
The challenge of deepfakes is real. But so is the need to protect open political discourse.
The question is not whether to act, but how.
Policies that are too broad, too vague, or too reliant on discretionary enforcement risk creating new problems while attempting to solve existing ones.
HB 5342 reflects that tension.
And as written, it may address the problem, but not in a way that provides clarity, predictability, or confidence in how it will be applied.
