State laws governing deepfakes are emerging quickly as policymakers respond to the risks posed by AI‑generated synthetic media—especially in elections, harassment, fraud, and public‑safety contexts. New Jersey, South Dakota, and Colorado have enacted some of the clearest statutory frameworks, each targeting a different aspect of deepfake misuse. Together, they illustrate how states are building a patchwork of rules around disclosure, prohibited uses, and penalties.
New Jersey – A350 (Deepfake Harassment and Non‑consensual Content)
New Jersey’s A350 focuses on non‑consensual deepfake sexual imagery and other forms of synthetic‑media harassment. While the bill has undergone multiple revisions across sessions, its core purpose is to criminalize the creation or distribution of sexually explicit deepfakes without the depicted person’s consent.
Key elements typically include:
- Criminal penalties for distributing sexually explicit deepfakes without consent.
- Civil remedies for victims, including damages and injunctive relief.
- Broader definitions of “digitally altered” or “synthetic” media to capture emerging AI tools.
This approach reflects a national trend toward treating deepfake sexual imagery as a form of digital sexual exploitation.
New Jersey – AR141 (Election‑Related Deepfake Resolution)
New Jersey’s AR141 is a legislative resolution rather than a binding statute, but it signals the state’s intent to regulate election‑related deepfakes.
Its themes include:
- Recognizing the threat deepfakes pose to election integrity.
- Calling for federal action or interstate cooperation.
- Encouraging the development of standards for disclosure and authentication of political media.
While not enforceable law, AR141 shows New Jersey’s movement toward a two‑track approach: criminalization of harmful deepfakes (A350) and election‑integrity protections (AR141).
South Dakota – SB164 (2025)
South Dakota’s SB164 is a fully enacted statute that directly targets deepfakes used to influence elections. It prohibits creating or distributing a deepfake intended to mislead voters about a candidate or election‑related conduct.
The law defines a deepfake as AI‑generated or digitally manipulated media that is so realistic a reasonable person would believe it depicts actual speech or conduct. It explicitly bans using such media to influence an election and establishes penalties for violations.
This makes South Dakota one of the first states to adopt a criminal prohibition on election‑related deepfakes.
Colorado – HB24‑1147 (Candidate Election Deepfake Disclosures)
Colorado’s HB24‑1147 creates a comprehensive disclosure regime for deepfakes in political communications.
Key requirements include:
- Mandatory disclosure when a communication contains a deepfake of a candidate.
- Prohibition on distributing undisclosed or improperly disclosed deepfakes with knowledge or reckless disregard of their falsity.
- Enforcement mechanisms and a private right of action for candidates harmed by deepfake misuse.
Colorado’s approach is notable for combining transparency, enforcement, and candidate‑initiated remedies, making it one of the strongest election‑deepfake laws in the country.
Cross‑State Themes
Across these laws, several regulatory patterns are emerging:
- Election integrity: South Dakota and Colorado focus on preventing deepfakes from misleading voters or impersonating candidates.
- Non‑consensual imagery: New Jersey’s A350 targets sexual deepfakes and harassment.
- Disclosure requirements: Colorado mandates clear labeling of political deepfakes; other states are considering similar rules.
- Criminal penalties: South Dakota imposes criminal sanctions for election‑related deepfake misuse.
- Civil remedies: Colorado and New Jersey provide private rights of action, reflecting a trend toward empowering victims directly.
These laws collectively show that states are regulating deepfakes through sector‑specific statutes—elections, harassment, consumer protection—rather than a single unified deepfake code