PPenwell Law

← Blog

Washington's SSB 5886 Is Now in Effect: Compliance Exposure for AI Voice Synthesis Platforms

Here is a hypothetical I expect to become very real for a lot of companies this year, put the way a board or a general counsel would put it:

Your company offers a real-time voice synthesis API used by podcast creators and social media platforms. Washington’s SSB 5886 took effect this week. Walk me through the compliance exposure and the first three steps you’d take as product counsel.

Here is how I would walk it.

Washington’s Substitute Senate Bill 5886 became effective on June 11, 2026, as Chapter 69, Laws of 2026. It amends the state’s Personality Rights Act, RCW 63.60, to add a new protected attribute, the “forged digital likeness,” alongside name, voice, signature, photograph, and likeness. The bill passed the Senate 47 to 0 on February 5, passed the House 85 to 9 on March 4, and was signed by Governor Bob Ferguson on March 16, 2026.

If your product generates synthetic audio derived from real human voices, read this one closely. The early coverage overstates the statute in some places and misses its sharpest edges in others. Mapping it wrong cuts both ways: you spend money overcorrecting, or you miss real exposure.

What SSB 5886 changes in Washington’s Personality Rights Act

The amended definitions section, RCW 63.60.020(3), defines a forged digital likeness as a visual representation or an audio recording, “either persistent or transmitted in real-time,” of an actual and identifiable individual, which (a) has been digitally created, adapted, altered, or modified to be indistinguishable from a genuine representation or recording of the individual, (b) misrepresents the appearance, speech, or conduct of the individual, and (c) is likely to deceive a reasonable person into believing the representation or recording is genuine.

Three features of that definition matter for a voice synthesis product.

The real-time language is the novel part. Most right-of-publicity statutes were drafted around fixed works: an advertisement, a recording, a printed image. Washington’s legislature deliberately extended the definition to representations transmitted in real time, and the legislative record is explicit about why: testimony before both chambers cited the rise of real-time cloning of likenesses in the commission of crimes over live video platforms. For an API that synthesizes voice output as a live stream rather than a stored file, this closes what might otherwise have been a colorable argument that the statute reaches only persistent recordings.

The test is conjunctive. All three elements must be satisfied: indistinguishability, misrepresentation, and likelihood of deceiving a reasonable person. Output that evokes a vocal style without being indistinguishable from a specific identifiable person should fall outside the forged digital likeness definition. Falling outside this definition does not end the analysis, a point I return to below.

The right extends to the deceased with the same remedies. The amended RCW 63.60.010 confirms the property right exists in the forged digital likeness of individuals and personalities deceased before, on, or after June 11, 1998, and the remedies provision applies its enhanced damages to infringement of “a living or deceased individual’s or personality’s forged digital likeness.” Under the pre-existing duration rules in RCW 63.60.040, the right lasts seventy-five years after death for personalities (persons whose identity attributes had commercial value) and ten years for other individuals. Washington also applies its statute regardless of place of domicile or place of domicile at time of death, a deliberate rejection of the usual domicile-at-death choice-of-law rule and one of the most aggressive extraterritoriality provisions in any state publicity statute.

What a plaintiff can recover under RCW 63.60.060

The amended RCW 63.60.060 provides (note that the official RCW pages may still display pre-amendment text pending codification; the amended language is in the enrolled bill):

  • Injunctive relief on reasonable terms.
  • A civil penalty of $3,000, replacing the prior “greater of $1,500 or actual damages” formula. The penalty is now cumulative with damages, not an alternative floor.
  • Actual damages sustained as a result of the infringement.
  • Disgorgement of profits attributable to the infringement and not already counted in actual damages, with the burden split: the injured party proves gross revenues attributable to the infringement, and the infringer proves deductible expenses.
  • Mandatory noneconomic damages for forged digital likeness claims. Where the infringement involves a forged digital likeness, the infringing party “shall also be liable for the noneconomic damages sustained as a result of the infringement whether or not the infringement resulted in any profit.” Washington law defines noneconomic damages to include subjective losses such as pain and suffering, mental anguish, injury to reputation, and humiliation (RCW 4.56.250), all recoverable here without any profit showing, and the operative word is “shall.”
  • Impoundment and destruction of infringing materials.
  • Attorneys’ fees, expenses, and costs to the prevailing party, which cuts both ways: a defendant who beats a weak claim can recover too.

Two details change how you model the risk. The statute creates a private right of action enforced in superior court. It contains no Attorney General enforcement authority, no civil investigative demand power, and no Consumer Protection Act hook. The enforcement risk is plaintiffs and the plaintiffs’ bar. Model it accordingly: mandatory noneconomic damages plus a fee-shifting clause is a structure designed to make individual claims economically viable for contingency counsel, which is its own kind of enforcement multiplier.

And on scope: the statute does not reach truly private, non-commercial uses. The operative infringement provision, RCW 63.60.050, still requires use on or in goods, merchandise, or products entered into commerce in Washington, or for advertising, fund-raising, or solicitation of donations. The frequently quoted line that infringement may occur “without regard to whether the use or activity is for profit or not for profit” means the statute reaches nonprofits and unmonetized commercial-context uses; it does not convert RCW 63.60 into a general anti-deepfake tort. Purely private malicious deepfakes are instead the territory of Washington’s companion criminal statute, Substitute House Bill 1205 (Chapter 51, Laws of 2025, effective July 27, 2025), which made knowing distribution of a forged digital likeness with intent to defraud, harass, threaten, or intimidate a gross misdemeanor.

Where tool-provider liability actually lives: “uses or authorizes”

Now for the part that actually decides a tool provider’s exposure. SSB 5886 contains no provision addressing the liability of the technology provider as such. Compare Tennessee’s ELVIS Act, which created express liability for anyone who makes available an algorithm, software, or tool whose “primary purpose or function” is producing a particular identifiable individual’s voice or likeness, with knowledge that the use was not authorized. Compare the proposed federal NO FAKES Act, which pairs a tool-liability standard with a DMCA-style notice-and-takedown safe harbor. Washington enacted neither the sword nor the shield.

What Washington’s statute says instead is that liability attaches to any person who “uses or authorizes the use” of the protected attribute in the covered commercial contexts. For a voice synthesis API company, the entire secondary-liability question compresses into one word: does providing the tool, on your terms, authorize your customer’s infringing use? No Washington court has construed that language as applied to an AI tool provider yet. That cuts two ways. You cannot rely on an intermediary safe harbor, because there is none; HB 1205’s criminal safe harbors for interactive computer services do not appear in the civil statute. And the things you can control, your terms, your consent architecture, your marketing, and your product design, are precisely the things a court would examine in deciding whether you authorized anything.

A note on jurisdiction, because the board will ask. A company with no Washington operations has real personal-jurisdiction defenses under the Ninth Circuit’s purposeful-direction cases, and the statute’s domicile-disregarding reach has its own constitutional exposure: in Experience Hendrix, L.L.C. v. HendrixLicensing.com, Ltd., 762 F.3d 829 (9th Cir. 2014), the Ninth Circuit upheld Washington’s extraterritorial approach only in narrow circumstances involving actual in-state harm, after the district court had found it unconstitutional. Those are litigation defenses, not compliance strategies. If Washington residents use your API or appear in its outputs, plan as though the statute applies.

The first three steps for product counsel

Step one: product inventory mapped to the statutory elements

Catalog every feature that generates audio derived from, conditioned on, or evocative of a real person’s voice, and sort the outputs against the actual definition rather than against marketing labels. True voice cloning of an identifiable person, where input voice maps to output identity, sits squarely inside the forged digital likeness definition and triggers the mandatory noneconomic damages regime, including for real-time streams. Fully synthetic voices not traceable to any actual individual sit outside it. Style transfer occupies the middle: output that is influenced by but not indistinguishable from a specific person likely fails the definition’s first element. But clearing the forged digital likeness definition does not clear the statute. The base statute has protected “voice” since 1998, and the common-law tradition running through Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988), teaches that a deliberately recognizable voice imitation used in advertising is actionable even without digital indistinguishability. Grade each feature against both the new definition and the old voice right.

Step two: consent audit against Washington’s standard

The statute requires “written or oral, express or implied consent of the owner of the right.” That is a forgiving formulation, more forgiving than the specific written-consent regimes some states apply to performers, and it means a well-built consent architecture actually does some work here. The audit has three layers. First, your own consent capture: any feature that clones an identifiable real person should require verifiable consent from that person, documented, not merely a checkbox from the customer. Second, your terms of service: customer representations and warranties that they hold all necessary consents, paired with indemnification, will not immunize you, but they are the raw material of any argument that you authorized only lawful uses. Third, the gap analysis: terms drafted against the older commercial-appropriation paradigm almost certainly never contemplated real-time transmission or postmortem rights reaching back decades. Find the gaps before opposing counsel does. This is the kind of platform-terms and content-rights work that should be revisited every time a statute like this lands.

Step three: monitoring and authorization hygiene

With no Attorney General to issue guidance, the law’s meaning will be developed by courts, so the monitoring program is litigation-focused: track filings in Washington superior courts and the Western District, and watch in particular for the first decisions construing “authorizes” against a tool provider and resolving whether the RCW 63.60.070 exemptions for news, commentary, satire, parody, and artistic works apply to forged digital likeness claims at all. SSB 5886 did not amend the exemptions section, which still lists only the five legacy attributes, an omission that practitioners, including Davis Wright Tremaine’s media group, have flagged as a serious open question with First Amendment stakes. Operationally, pair the monitoring with the controls that make your authorization story credible: abuse detection, takedown and termination tooling, and provenance tracking for generated outputs.

On provenance, watch what the music industry just did. On June 10, 2026, Warner Music Group announced an agreement to acquire Sureel AI, the Palo Alto attribution startup whose registry tracks how AI models use creative works and whose name-image-likeness suite traces voice clones and style replication in AI outputs. Sureel is a music-industry provenance tool, not a turnkey compliance answer for a voice API. But per-output attribution of this kind is the architecture that consent-verified voice products will increasingly be measured against, and it is worth understanding now.

How SSB 5886 compares: the ELVIS Act, California, New York, and the NO FAKES Act

Washington is the latest move in a patchwork that keeps tightening. Tennessee’s ELVIS Act has covered voice simulation, with tool-provider liability, since July 2024. California’s AB 1836 created a digital-replica right for deceased performers, and AB 2602 voids inadequately specified replica licenses for living performers, both effective January 2025. New York amended its postmortem performer statute in December 2025. And on May 20, 2026, a bipartisan group reintroduced the federal NO FAKES Act (S. 4591 and H.R. 8915) with a new counter-notification procedure; it has broad industry support but, as of this writing, remains a bill, not a law. If it passes, its notice-and-takedown safe harbor would partially reorder the analysis above.

Until then, Washington’s combination of real-time coverage, universal application regardless of fame or domicile, mandatory emotional-distress damages, and fee shifting makes it the statute your product and consent architecture should be built to satisfy. A product that clears Washington clears most of the field.

Frequently asked questions

When did Washington’s SSB 5886 take effect? June 11, 2026, ninety days after the legislature adjourned, as Chapter 69, Laws of 2026.

Does SSB 5886 apply to AI voice cloning companies outside Washington? The statute applies to covered uses in Washington commerce regardless of where the rights holder lived or died, and Washington courts can reach out-of-state defendants whose conduct is purposefully directed at the state. Jurisdictional defenses exist, but if Washington residents use your product or appear in its outputs, plan as though the statute applies.

Does SSB 5886 cover voices of deceased people? Yes. The forged digital likeness right extends to deceased individuals and personalities with the same remedies, for ten years after death for individuals and seventy-five years for personalities whose identity had commercial value.


I serve as outside product, commercial, and IP counsel to creative-technology companies across the product lifecycle: IP and content rights, copyright and training-data questions, music and entertainment licensing, platform terms, privacy, and AI governance, including the right-of-publicity issues that statutes like SSB 5886 create. If this touches your product, get in touch.

This article is general information, not legal advice, and should not be relied on as such; it reflects the law as of June 12, 2026, and reading it does not create an attorney-client relationship. Liam J Penwell is the professional name of Jason Penwell, Esq. (State Bar of California No. 339157), licensed in California. Penwell Law · Burlingame, California · Contact.

← Back to all posts

Subscribe

New posts, in your inbox.

About one email a week, whenever something new goes up: AI governance, privacy, and IP & content-rights commentary. No spam; unsubscribe in one click.

We’ll only use your email to send new posts. Unsubscribe anytime. See our Privacy Policy.