The National Association of Voice Actors (NAVA) brought its advocacy to Washington, D.C. this week, with co-founders Tim Friedlander and Carin Gilfry appearing on panels examining how artificial intelligence is transforming creative work. The events, running May 12 and 13, span the DC Commission on the Arts and Humanities and the Martin Luther King Jr. Memorial Library.
The timing is not accidental. A new survey from NAVA found that 21 percent of voice actors reported losing work directly to AI in 2026 — up from 14 percent just one year ago. That is a seven-percentage-point jump in twelve months, and it is accelerating.
What NAVA Is Pushing For
NAVA is supporting two pieces of federal legislation making their way through Congress. The NO FAKES Act in the Senate would establish a federal right of publicity. The NO AI Frauds Act in the House goes further, explicitly recognising “voice” as a protected category and a property right. NAVA has also endorsed the Generative AI Copyright Disclosure Act, which would require AI companies to disclose when copyrighted works are used to train their models.
Taken together, these bills represent the first serious attempt to define what AI can and cannot do with a person’s voice at the federal level. Some of this is already being handled at the state level — Washington State, Tennessee, and others have passed laws governing AI-generated likenesses and voices — but the absence of a unified federal standard has created uncertainty for businesses deploying voice AI at scale.
The advocacy groups’ argument is straightforward: voice is property. If an AI company trains a model on a professional voice actor’s work and then uses that model to generate new audio, they should need permission and should pay for it. That principle, if codified in law, would have significant implications across the voice AI industry.
The Numbers That Matter
The 21 percent figure deserves some context. NAVA represents professional voice actors — people who earn their living from commercial voiceover, audiobooks, animation, gaming, and corporate narration. These are not hobbyists. When 21 percent of them report losing paid work to AI-generated audio, it reflects real displacement in a specific skilled labour market.
The seven-point jump in one year tracks with what has happened in the tools available to businesses. In 2024, generating high-quality synthetic voice required technical expertise and significant cost. By 2026, the barrier is dramatically lower. Several platforms now offer voice cloning in under 30 seconds from a short audio sample. The creative economy felt that shift first because it is where the demand for professional audio production concentrates.
The broader voice AI market is growing fast regardless. Voice AI investment jumped from roughly $315 million in 2022 to $2.1 billion in 2024. The global voice recognition market is expected to be worth $22 billion in 2026 and is projected to nearly triple over the next five years. The legislative push from NAVA is happening inside an industry that is expanding rapidly, not contracting.
What This Means for Business
This story has two distinct layers, and most businesses are conflating them in ways that will cause problems later.
Layer one is voice cloning. If you are replicating a specific person’s voice — a celebrity, a professional voice actor, an internal employee — without their consent, you are in legally uncertain territory right now and increasingly likely to face legal exposure as federal legislation advances. The NO AI Frauds Act, if passed, would make this a federal property rights violation. State laws in Tennessee and Washington already provide some protection. If your marketing strategy involves generating synthetic audio that sounds like a real individual, you need legal counsel, not just a terms-of-service checkbox.
Layer two is enterprise voice AI. This is categorically different. Deploying a voice AI employee to handle inbound calls, answer customer questions, or surface information from internal knowledge bases does not involve cloning anyone’s voice. The AI uses a synthetic voice with no connection to any identifiable human. This is the use case that enterprise voice AI platforms are built around, and it sits well outside the legislative concerns NAVA is raising.
The confusion between these two things is creating a generalised hesitancy about voice AI that is not warranted for businesses pursuing legitimate operational use cases. A voice AI employee that handles after-hours calls or routes internal support requests is solving a workflow problem, not a copyright problem.
What businesses should actually watch is whether the Generative AI Copyright Disclosure Act passes, which would affect how they evaluate AI vendors. If a voice AI platform cannot explain whether its underlying models were trained on licensed or public-domain audio, that is a question worth asking now, before it becomes a due diligence requirement.
The practical read: The regulatory pressure building around voice AI is targeted at synthetic voice cloning of real individuals — not at enterprise automation. If you are deploying voice AI for operational reasons rather than creative voice replication, the regulatory trajectory gives you more clarity, not less. The same legislation that protects voice actors from unauthorised cloning also de-risks legitimate enterprise use by distinguishing it clearly from the practices under scrutiny.
The businesses that will navigate this well are the ones that understand what type of voice AI they are deploying and can explain the distinction clearly. The ones that lump all voice AI together will either delay adoption unnecessarily or expose themselves to risk they did not bother to understand.
The practical next step is the free Working With Claude field guide. Thirty-two pages covering the ecosystem, Claude Code, and how to govern a rollout properly. Get your copy.