Voice Changer Impersonation Laws by Country

Voice impersonation law is evolving fast — and if you use a voice changer, AI voice tools, or create audio content featuring other people’s voices, you need to understand where the legal lines are in your country. This guide covers the US, EU, Brazil, and UK: what is prosecuted, what is protected, and what the 2024–2026 legal changes mean for creators, gamers, and developers.

This is not legal advice. For specific situations, consult a qualified attorney in your jurisdiction.

TL;DR

Voice changers are legal tools; criminal liability depends on intent and context, not the software itself.
The US has a patchwork of state and federal laws — Tennessee’s ELVIS Act and California’s Right of Publicity are the most relevant for AI voice.
The EU AI Act (2026) mandates transparency labeling for synthetic audio and treats voice as biometric data under GDPR.
Brazil applies LGPD (voice = biometric data), Marco Civil, and Código Penal Art. 307 to voice-based impersonation.
The UK combines the Online Safety Act 2023 with defamation law — non-consensual intimate deepfakes became a specific criminal offense in 2024.
Parody, fiction, journalism, and education are protected in all four jurisdictions — with important nuances.

What “Voice Impersonation” Means Legally

Voice impersonation law does not exist as a single unified category in most countries. Instead, it sits at the intersection of several legal frameworks: identity fraud, defamation, personality rights, data protection, and (increasingly) AI-specific regulation.

The core legal question is almost always the same: did the impersonation cause harm, was consent given, and was there deceptive intent? A comedian doing a pitch-perfect impression of a politician on stage is protected expression. The same voice clip embedded in a fake news video presented as genuine is a different situation entirely.

The introduction of AI voice tools has sharpened these questions because the barrier to convincing impersonation dropped from “professional impression artist” to “anyone with a laptop.” Courts and legislators in every major jurisdiction are still catching up.

United States: State-Level Patchwork

Federal law in the US covers voice impersonation primarily through wire fraud (18 U.S.C. § 1343), the TRACED Act for phone-based spoofing, and the FTC Act for consumer deception. The more active legal development has happened at the state level.

Tennessee: The ELVIS Act (2024)

Tennessee passed the Ensuring Likeness Voice and Image Security (ELVIS) Act in March 2024, the first US state law explicitly protecting an individual’s voice from unauthorized AI replication. It extends the state’s existing personality rights statute to cover voice specifically.

Key provisions:

Prohibits using AI to produce a “sound-alike” of a person’s voice for commercial purposes without consent
Covers both living and deceased individuals
Applies to distribution platforms that knowingly host violating content
Does not require that the voice be famous — it applies to any individual

The ELVIS Act was motivated by concerns in the music industry but applies broadly. A streamer creating AI voice content of a private individual for commercial gain could face liability in Tennessee courts if the content is distributed there.

California: Right of Publicity (AB 2602 and SB 1073, 2024)

California had existing right-of-publicity law covering name, voice, and likeness. In 2024, two bills expanded these protections to AI contexts:

AB 2602 requires explicit written consent in contracts before an AI can be used to create a digital replica of a performer’s voice or likeness for work they could have been hired to do.
SB 1073 extends the right-of-publicity protections to deceased personalities and clarifies that AI-generated voice replicas require the same consent as a live performance.

California’s law is particularly significant because the entertainment industry is concentrated there and the state’s court decisions often influence national standards.

New York: Civil Rights Law §§ 50–51

New York’s Civil Rights Law Sections 50–51 have long prohibited commercial use of a person’s name, portrait, picture, or voice without written consent. The statute predates AI but courts have applied it to digital replicas. NY Senate Bill S8641 (introduced 2024) would explicitly add AI-generated voice and likeness — as of 2026 it has passed committee.

The New York framework matters because it allows injunctive relief and actual damages, making it a practical enforcement tool beyond just criminal law.

Federal: What Exists and What Is Pending

At the federal level, the NO FAKES Act (proposed 2023, reintroduced 2024) would create a federal right against unauthorized AI replicas of voice and likeness, superseding the state patchwork with a uniform standard. As of May 2026, it has not passed, but bipartisan support means passage is likely within the next legislative cycle.

Existing federal hooks for voice impersonation cases include:

Wire fraud (18 U.S.C. § 1343): Any scheme to defraud using wire communication, including voice. Used in deepfake phone scams.
Computer Fraud and Abuse Act: When voice manipulation is part of unauthorized computer access.
FTC consumer protection rules: FTC issued a ruling in 2024 explicitly covering AI-generated voice cloning in telemarketing fraud.

What Is Protected in the US

Parody and satire: First Amendment protects obvious parody; the content must be recognizable as commentary, not mistakable for genuine speech.
Fiction and entertainment: Dramatizations, movies, and fictional audio content using voice impressions are protected as creative expression.
Journalism: Reporting on public figures, including audio analysis and demonstration, falls under press freedom protections.
Education: Academic study of voice technology, demonstrations for pedagogical purposes.
Personal use: Changing your own voice for privacy, fun, or gaming — no victims, no liability.

The EU has taken a regulation-first approach, building AI voice rules on top of an existing strong data-protection framework.

Under GDPR Article 9, voice recordings that allow identification of a natural person are classified as biometric data — a special category requiring explicit consent for collection and processing. This applies even to short audio clips.

Practical implications for voice changer users:

Recording and publishing someone’s voice without consent for identification purposes is a GDPR violation
Training AI voice models on someone’s voice data without consent is a serious breach, subject to fines of up to 4% of global annual turnover for companies or significant fines for individuals
“Legitimate interest” does not apply to biometric data — only explicit consent or specific legal bases (law enforcement, public health)

EU AI Act: Transparency Obligations (2026)

The EU AI Act entered its full enforcement phase in 2026. For audio content, the key provision is Article 50, which requires providers of AI systems that generate synthetic audio (including voice) to ensure outputs are labeled as AI-generated. This is the “watermarking” or transparency obligation.

What this means in practice:

AI-generated voice content distributed publicly must carry a machine-readable label indicating it is synthetic
Platforms that host synthetic audio at scale must implement detection and labeling
Deep fake audio that could deceive the public is classified as a prohibited AI practice under Article 5 when it causes harm

The EU AI Act also explicitly prohibits AI systems used for subliminal manipulation and exploitation of vulnerabilities — both of which could apply to persuasive synthetic voice content designed to deceive.

Member State Criminal Law

Beyond EU-level regulation, member states have their own criminal law that applies to voice impersonation:

Fraud: Creating a synthetic voice clip to deceive someone into financial transactions is fraud in every EU member state, regardless of the AI tool used.
Defamation: Generating a false audio clip attributed to a real person that damages their reputation is defamation — a civil or criminal offense depending on jurisdiction.
Sexual deepfakes: Multiple EU member states criminalized non-consensual intimate synthetic media in 2023–2024. Germany, France, and Spain all have explicit provisions.

Country Comparison: EU

Member State	Key Provision	Notable Enforcement
Germany	§201a StGB (violation of intimate sphere), §238 StGB (stalking)	Criminal prosecution for intimate deepfakes
France	Art. 226-1 CP (privacy violation)	Active enforcement against revenge deepfakes
Spain	Ley Orgánica 3/2018 (data protection + digital rights)	Right to digital oblivion extends to synthetic voice
Netherlands	GDPR implementation particularly strict	DPA fines for biometric processing without consent

United Kingdom: Online Safety Act + Defamation

The UK post-Brexit operates its own legal framework, though it shares roots with EU law and has moved quickly on synthetic media.

Online Safety Act 2023

The Online Safety Act 2023 made sharing non-consensual intimate deepfakes a criminal offense in England and Wales. This directly covers AI-generated intimate audio content of a real person distributed without consent. The offense carries up to two years imprisonment.

Key provisions relevant to voice:

Platforms have a duty of care to remove deepfake content — including audio
Creating intimate synthetic content of a person without consent, with intent to cause distress, is criminal even if not shared
Ofcom (the regulator) can fine platforms up to 10% of global revenue for non-compliance

Defamation Act 2013

The Defamation Act 2013 provides a framework for claims involving false statements that damage reputation. A realistic synthetic audio clip that falsely attributes statements to a person and is published to a third party is a strong defamation candidate.

The UK standard requires showing “serious harm” to reputation — a higher bar than some US states but consistently met by convincing deepfake audio cases.

Malicious Communications Act 1988 / Communications Act 2003

Both acts cover electronic communications sent to cause distress or that are grossly offensive. Synthetic voice messages designed to harass an individual can be prosecuted under these statutes, independent of whether the impersonation reaches the defamation threshold.

What Is Protected in the UK

Public interest journalism: Privilege applies to reporting on matters of public interest, including demonstrating voice technology for investigative purposes.
Satire and commentary: UK courts recognize creative satirical content, though the parody defense in copyright law (Section 31B CDPA 1988) is narrower than US First Amendment protections.
Academic research: Processing voice data for genuine research purposes under legitimate academic frameworks is protected.
Personal entertainment: Changing your own voice, creating fictional content, gaming personas — all legal absent harm or deception.

Brazil: LGPD + Marco Civil + Código Penal

Brazil does not yet have legislation specifically targeting AI-generated voice, but the intersection of three existing legal frameworks creates meaningful protection — and real criminal exposure.

LGPD: Lei Geral de Proteção de Dados

The LGPD (Law 13.709/2018) is Brazil’s data protection law, modeled closely on GDPR. Voice data that allows identification of a natural person is classified as biometric data under Article 5, Item II — a sensitive category requiring explicit, informed consent for processing.

Relevant implications:

Collecting, storing, or processing someone’s voice data without consent violates the LGPD
Companies or individuals using voice data for AI training without consent face fines from the ANPD (Autoridade Nacional de Proteção de Dados)
The TSE (Tribunal Superior Eleitoral) applied LGPD principles aggressively during the 2022 and 2024 electoral cycles to combat synthetic audio disinformation

Marco Civil da Internet (Law 12.965/2014)

Marco Civil establishes a Bill of Rights framework for internet use in Brazil. Key provisions for voice content:

Article 19: Platforms are generally not liable for third-party content unless they fail to comply with a specific court order to remove it — the “notice and takedown” model
Article 21: Platforms have stricter liability for non-consensual intimate content, requiring immediate removal upon notification (a model later exported to the EU’s DSA)

For creators: publishing synthetic voice content of another person that causes harm triggers Article 21 takedown rights. Platforms that do not comply become jointly liable.

Código Penal: Falsa Identidade (Art. 307)

Article 307 of the Brazilian Código Penal criminalizes assuming a false identity to obtain an advantage or cause harm to another person. The maximum penalty is one year imprisonment or fine.

While Article 307 predates voice synthesis technology by decades, prosecutors have begun applying it to deepfake audio cases — particularly in fraud scenarios where someone uses a synthetic voice to impersonate a family member or business contact in financial conversations (the “voice scam” / “golpe do WhatsApp de voz”).

Additional criminal hooks:

Art. 171 CP (Estelionato / Fraud): Using voice impersonation as part of a scheme to obtain financial advantage
Art. 139 CP (Difamação) and Art. 140 CP (Injúria): Defamation and injury to honor — applicable to false audio clips spreading damaging claims
Electoral Law (Law 9.504/1997, Art. 323): Specific prohibition on creating and distributing false content affecting elections, actively enforced by the TSE for synthetic audio during campaign periods

Electoral Enforcement in Brazil

Brazil has been particularly active on electoral deepfake enforcement. The TSE in 2022–2024 issued multiple resolutions specifically addressing AI-generated audio and video in campaigns:

Synthetic media in electoral advertising requires explicit disclosure
Manipulated audio designed to damage a candidate’s reputation is prosecutable under electoral law
Content removed from platforms must stay down during the electoral period

Country-by-Country Comparison Table

Jurisdiction	Voice as Protected Data	Primary Criminal Risk	AI-Specific Law	Parody Protected
United States (Federal)	Limited (no federal data law)	Wire fraud, CFAA	NO FAKES Act (pending)	Yes — First Amendment
Tennessee	ELVIS Act (voice right)	Unauthorized commercial AI voice	ELVIS Act (2024)	Yes, with limits
California	Right of Publicity	Unauthorized AI performer replica	AB 2602, SB 1073 (2024)	Yes, transformative use test
New York	Civil Rights §§ 50-51	Commercial use without consent	SB 8641 (pending)	Yes
EU (all states)	GDPR Art. 9 — biometric data	Fraud, defamation, sexual deepfakes	EU AI Act Art. 50 (2026)	Yes, narrower defense
Germany	GDPR + §201a StGB	Intimate deepfakes, stalking	EU AI Act	Yes
UK	UK GDPR	Online Safety Act — intimate deepfakes	OSA 2023	Yes, satire defense
Brazil	LGPD Art. 5 — biometric data	Art. 307 CP (falsa identidade), fraud	None specific yet	Yes, limited

What Is Prosecuted: The High-Risk Scenarios

Across all four jurisdictions, these scenarios consistently trigger criminal or civil liability:

Financial fraud using synthetic voice: Cloning someone’s voice to trick a family member, colleague, or business into transferring money. This is prosecuted as fraud regardless of jurisdiction. Several high-profile cases in 2023–2025 involved WhatsApp voice message scams where victims wired money believing they were speaking with a family member.

Non-consensual intimate deepfakes: Creating or distributing realistic sexual audio or video of a person without consent. Criminal in the UK (OSA 2023), most EU member states, and several US states. Civil liability exists broadly.

Electoral disinformation: Synthetic audio designed to damage a political candidate or spread false statements attributed to them. Actively prosecuted in Brazil; covered under election integrity laws in the EU; addressed by the FEC in the US.

Defamation: Realistic audio clips that falsely attribute harmful or criminal statements to a real, identifiable person. Civil liability everywhere; criminal defamation in some jurisdictions (Brazil, some EU states).

Identity fraud / account takeover: Using a voice clone to bypass voice-based authentication systems. Fraud and computer access offenses apply in all jurisdictions.

What Is Legal: The Protected Use Cases

Understanding where protection exists is as important as knowing the risks.

Parody and satire: Creating obviously comedic content using a voice impression or synthetic voice, clearly labeled as parody, is protected expression in all four jurisdictions. The key tests: is it recognizable as parody, and could a reasonable person be deceived into thinking it is genuine?

Fiction and creative work: A novel, film, game, or audio drama can feature characters with synthesized voices — including voices based on public figures — when the fictional nature is clear. Historical fiction using AI voice reconstruction of public figures for educational media occupies a gray zone but is generally defensible.

Journalism and investigation: Demonstrating how voice synthesis technology works, including generating examples, is protected press activity. Investigative journalism using synthetic voice reconstruction to analyze what was said in degraded recordings has been upheld.

Education and research: Academic use of voice synthesis for research, including generating training data with consenting speakers, is protected. Teaching about deepfake detection and voice synthesis technology is fully protected.

Personal use and gaming: Changing your own voice for online gaming, privacy in calls, or creative streaming content does not implicate any of these laws. Voice changers as a category are neutral tools — the legal question always attaches to harm and deception, not the technology itself.

For a deeper look at the ethical dimensions beyond the legal minimums, see our post on ai-voice-generator-celebrity-ethics.

The Transparency Obligation: Labeling Synthetic Audio

A theme emerging across all jurisdictions is the mandatory disclosure of AI-generated content. The EU AI Act makes this legally required for synthetic audio; the UK Online Safety Act and US FTC guidance point in the same direction.

For creators using voice changer or AI voice tools:

Label AI-generated voice content clearly in video descriptions, posts, or transcripts when publishing publicly
Do not present synthetic voice as genuine recordings of real people without disclosure
If content features a synthetic version of another person’s voice, disclose this in the content itself

This is both a legal risk-reduction measure and an ethical standard that protects your audience’s ability to assess what they are hearing.

Practical Guidance for Different Use Cases

Streamers and content creators: Using a voice changer for your gaming persona, applying effects to your own voice, or creating fictional characters has zero legal exposure. Creating content that mimics a real person in ways that could be mistaken for genuine audio, without disclosure, increases risk significantly as laws tighten.

Developers building voice tools: The GDPR / LGPD biometric data classification means that if you collect users’ voice data for training, explicit informed consent is mandatory. Privacy policies must disclose voice data use. EU users have deletion rights — you must be able to purge their voice data from models.

Businesses using AI voice in customer service: Requires disclosure to callers that they are interacting with a synthetic voice (already required in several US states and under the EU AI Act). Impersonating a human agent using AI voice without disclosure is prohibited.

Researchers and journalists: Broad protection exists, but document your legitimate purpose. If you are demonstrating a technology’s risks, label your examples clearly.

How This Affects Your Use of VoxBooster

VoxBooster’s voice changer and AI voice features are neutral tools — the same technology that lets you protect your privacy online, create a gaming persona, or produce creative audio content. The legal frameworks above all operate on intent and use, not on the existence of the software.

Using VoxBooster for your own voice transformations, creative content, streaming, and entertainment falls clearly within the protected categories in every jurisdiction discussed here. The risk scenarios — fraud, non-consensual intimate content, electoral manipulation — require a deliberately harmful intent that is entirely separate from normal use.

For context on ethical use of AI voice beyond the legal minimums, see voice-cloning-memorial-ethics for an exploration of the harder questions around consent and representation.

The broader legal and ethical framework for AI-generated content — and a checklist for staying clearly within it — is covered in voice-cloning-consent-legal-checklist. For an overview of where the ethics and law intersect in 2026, see voice-cloning-ethics-2026 and voice-cloning-deepfake-detection.

Frequently Asked Questions

Is using a voice changer illegal?

Using a voice changer is legal in most jurisdictions for entertainment, gaming, creative content, and privacy protection. It becomes illegal when combined with intent to deceive for fraud, impersonate someone to cause harm, create non-consensual intimate deepfakes, or spread electoral disinformation. The tool itself is neutral; the intent and context determine legality.

Can I legally impersonate a celebrity’s voice online?

Parody and satire with clearly labeled content is generally protected in the US, EU, and UK. However, realistic impersonation that implies endorsement, spreads false information, or constitutes defamation crosses legal lines. Tennessee’s ELVIS Act (2024) and California AB 2602 specifically protect performers from unauthorized AI voice replication for commercial purposes.

What is voice impersonation law in Brazil?

Brazil does not yet have a dedicated voice impersonation statute, but several laws apply: LGPD treats voice as biometric data requiring consent, Marco Civil da Internet establishes liability for harmful content, and Código Penal Article 307 criminalizes false identity assumption. The 2024 electoral period saw active enforcement of deepfake audio rules by the TSE.

Yes. Under GDPR Article 9, voice recordings that can identify a person are classified as biometric data — a special category requiring explicit consent for processing. The EU AI Act (in force from 2026) adds mandatory transparency requirements for AI-generated audio content, including disclosure when voice is synthetically altered or cloned.

What makes AI voice cloning illegal vs. legal?

AI voice cloning is legal for personal use, entertainment, creative fiction, journalism, and education when content is properly labeled and no harm is intended. It becomes illegal when used to impersonate someone for financial fraud, create non-consensual sexual content, defame a person, or interfere with elections — all areas where existing criminal and civil law applies regardless of the technology used.

Can I use a voice changer on phone calls legally?

In the US, using a voice changer on calls is generally legal for privacy or fun, but becomes illegal under the TRACED Act and wire fraud statutes if used to deceive someone for financial gain or to harass. Several states (California, Illinois) have specific anti-robocall and call-spoofing laws. In the EU and UK, consent-based recording and anti-fraud statutes apply.

Is voice cloning for parody protected under free speech?

In the US, parody is protected under the First Amendment when it is clearly recognizable as parody and not likely to be mistaken for genuine speech. The EU and UK have narrower parody defenses in IP law but similar free speech protections. The key test across jurisdictions: could a reasonable person believe the impersonation is real, and does it cause harm?

Conclusion

Voice impersonation law in 2026 is a patchwork — but the direction of travel is consistent across the US, EU, UK, and Brazil. Laws are tightening around consent, transparency, and harm. The use cases getting regulated are narrow and clearly harmful: fraud, non-consensual sexual content, defamation, election interference. The uses being protected are equally clear: parody, fiction, journalism, education, and personal expression.

If you use a voice changer for gaming, streaming, or creative content, none of this represents a practical legal risk for your normal activity. Understanding the framework helps you make good decisions at the margins — knowing to label AI-generated content, to avoid realistic impersonation of identifiable people without context, and to stay clear of the specific high-risk scenarios courts and legislators are targeting.

Voice changer legal risk comes from intent and harm, not from the technology. Keep your use in the entertainment and creative space — and label synthetic content clearly when it involves real people — and the law across all four jurisdictions leaves you substantial room to create.

Download VoxBooster — free 3-day trial, no credit card required.

This post is for informational purposes only and does not constitute legal advice. Laws vary by jurisdiction and change frequently. Consult a qualified attorney for guidance on specific situations.