Deepfake Law in the United States
The US now has a federal NCII deepfake law through the TAKE IT DOWN Act, while broader regulation remains fragmented across state laws, FCC robocall rules, and existing fraud statutes.
- Status
- enacted
- Jurisdiction
- United States
- Effective
- May 2025
- Statute
- TAKE IT DOWN Act + FCC TCPA + state deepfake laws
The United States still has no comprehensive federal deepfake law, but it no longer has a wholly state-by-state framework for the most common abuse category. The federal TAKE IT DOWN Act now covers non-consensual intimate imagery, including AI-generated intimate images, while broader deepfake regulation remains layered across TCPA robocall enforcement, wire fraud, CFAA, FTC action, and state laws.
Federal landscape
TAKE IT DOWN Act (Public Law 119-12). Enacted May 19, 2025. Creates a federal criminal prohibition on publishing or threatening to publish non-consensual intimate visual depictions, including realistic AI-generated intimate images, and requires covered online platforms to remove reported material within 48 hours after a valid notice. This is the most important federal deepfake-specific law now in force.
FCC (Federal Communications Commission). In February 2024, following the Biden robocall incident in New Hampshire, the FCC ruled that AI-generated voices in robocalls are illegal under the existing Telephone Consumer Protection Act. Penalties can reach millions of dollars for operators.
DEFIANCE Act. Passed the Senate in the 118th Congress in July 2024 but did not become the governing federal baseline. Its civil-cause-of-action approach remains influential in 2026 proposals, while TAKE IT DOWN supplies the operative federal criminal and notice-removal framework for NCII.
Existing fraud statutes. Wire fraud (18 U.S.C. § 1343), identity theft (18 U.S.C. § 1028A), and securities fraud already apply to deepfake-enabled scams. The Arup-style CEO fraud (case study) is prosecutable under existing law if perpetrators are caught and within jurisdiction.
FTC. Has pursued deepfake advertising scams under unfair and deceptive practices rules. Proposed AI-endorsement rules went effective in 2024.
State landscape (30+ election-deepfake states as of 2026)
Most state deepfake laws focus on two categories:
Non-consensual intimate imagery. Texas, Virginia, New York, California, Minnesota, Georgia, Illinois, and many more criminalize creation or distribution of AI-generated sexually explicit imagery without consent. State laws still matter because they can add civil remedies, broader creation offenses, and state criminal penalties beyond the federal platform-takedown structure.
Election interference. California, Texas, Michigan, Washington, Minnesota, New Mexico, and many others restrict deepfakes depicting candidates within specific pre-election windows, with varying disclaimer, takedown, and enforcement mechanisms. Campaign compliance is still state-by-state.
Penalties vary: California's election-deepfake law carries civil penalties and injunctive relief. Texas makes election deepfakes a criminal offense. Several states (Tennessee's ELVIS Act) extend right-of-publicity protections to cover AI voice clones.
Practical implications
For US-based organizations:
- Anyone operating robocalls: AI voices in robocalls are illegal under the TCPA. Mandatory screening at the carrier level is now standard.
- Platforms hosting user content: TAKE IT DOWN notice-and-removal obligations apply federally; state NCII laws and civil claims can layer on top.
- Political campaigns: state-by-state compliance review is required, and "safe-harbor" labeling conventions vary.
- Financial institutions: CEO-fraud by deepfake is prosecutable under existing law but detection-and-prevention is your responsibility — regulators expect it.
What's pending
- Federal publicity-right / voice-and-likeness bills, including NO FAKES-style proposals.
- Broader federal AI disclosure requirements following the EU AI Act model; still debated, not enacted as a horizontal deepfake-labeling rule as of July 2026.