Synthetic Media and Deepfakes Under Texas Law

Short Answer

Texas attacks harmful synthetic media on three fronts. Penal Code § 21.165, rewritten by two 2025 bills, criminalizes knowingly producing or distributing sexually explicit deepfake media of a real person without effective consent, with a separate offense for threats and narrow affirmative defenses for conduits and AI providers that meet written statutory conditions. Election Code § 255.004(d) makes it a Class A misdemeanor to create a deepfake video and cause its publication or distribution within 30 days of an election with intent to injure a candidate or influence the result. And the Securing Children Online through Parental Empowerment Act (SCOPE Act), Business and Commerce Code chapter 509, imposes age-registration, data-limit, parental-tool, and harmful-material duties on digital service providers serving known minors, enforced by the Attorney General, with a parent or guardian able to seek only a declaratory judgment or injunction. The Fifth Circuit has kept a preliminary injunction in place against the chapter's monitoring and filtering requirement while the rest of the chapter stands.

Which Laws Apply

Texas AI-specific criminal law: Penal Code § 21.165 as amended by HB 449 (89th Legislature, 2025) and SB 441 (89th Legislature, 2025).

Texas election law: Election Code § 255.004(d) and (e) (AI-specific provision within the generally applicable Election Code), added by SB 751 (86th Legislature, 2019); subsection (b) of the same section was held facially unconstitutional in Ex parte Stafford, a holding the Court of Criminal Appeals affirmed in 2024 without disturbing subsection (d).

Texas platform duties: Business and Commerce Code chapter 509 (the SCOPE Act, HB 18 (88th Legislature, 2023)).

Related coverage on this site: the guide "Deepfakes, Voice Clones, and Likeness" treats the civil track, including Civil Practice and Remedies Code chapter 98B claims against producers, nudification apps, and payment processors, the 72-hour site removal duty, and the 48-hour duty under the federal Tools to Address Known Exploitation by Immobilizing Technological Deepfakes on Websites and Networks Act (TAKE IT DOWN Act).

The Criminal Statute: Penal Code § 21.165

Two 2025 bills rewrote this section, and both took effect September 1, 2025. HB 449 changed "deep fake video" to "deep fake media," so the statute now covers any visual depiction, not just video. SB 441 made the broader rewrite, adding the threat offense, the consent definition, the affirmative defenses, and restitution.

"Deep fake media" means a visual depiction created or altered through software, machine learning, artificial intelligence, or any other computer-generated or technological means, including manual or automated adaptation of an authentic depiction, that appears to a reasonable person to depict a real person, indistinguishable from an authentic depiction of that person, performing an action that did not occur in reality. "Visual depiction" includes a photograph, motion picture film, videotape, digital image or video, or other visual recording. SB 441, Section 2

A person commits an offense if, without the effective consent of the person appearing to be depicted, the person knowingly produces or distributes by electronic means deep fake media that appears to depict the person with visible computer-generated intimate parts, with another human being's visible intimate parts presented as the person's, or engaging in sexual conduct in which the person did not engage. Consent counts only if the depicted person knowingly and voluntarily signed a written agreement drafted in plain language that includes a general description of the deep fake media and, if applicable, the audiovisual work into which it will be incorporated. SB 441, Sections 3(b), 3(b-2)

The offense is a Class A misdemeanor. It becomes a third-degree felony only if it is shown at trial that the actor was previously convicted of an offense under this section or that the person appearing to be depicted is younger than 18. The statute creates no other enhancements. SB 441, Section 3(c)

Threatening to produce or distribute deep fake media with the intent to coerce, extort, harass, or intimidate another person is a separate offense, a Class B misdemeanor, rising to a Class A misdemeanor for a repeat offender or a threat involving a depiction of a person younger than 18. A disclaimer stating the media was unauthorized, or a label indicating the depiction is not authentic, is not a defense. A court must order a convicted defendant to make restitution to the victim for psychological, financial, or reputational harm resulting from the offense. SB 441, Sections 3(b-1), 3(c-1), 3(c-2), 3(e)

The statute provides two affirmative defenses for intermediaries, written in precise terms. It is an affirmative defense that the actor is an Internet service provider, cloud service provider, cybersecurity service provider, communication service provider, or telecommunications network that transmits data, and acted solely in a technical, automatic, or intermediate nature. It is separately an affirmative defense that the actor is a provider or developer of a publicly accessible artificial intelligence application or software used to create the media, included a prohibition against creating media prohibited by the section in terms and conditions or user policies that a user must acknowledge before gaining access, and took affirmative steps to prevent its creation through technological tools, such as training the system to identify prohibited media, providing effective reporting tools, filtering prohibited media before it is shown to a user, and filtering prohibited media from training data. The statute does not add a condition that the defense is lost for failure to remove media upon notice. SB 441, Sections 3(c-4), 3(c-5)

Election Deepfakes: Election Code § 255.004(d)

Since 2019, Texas has criminalized election deepfakes. SB 751 (86th Legislature, 2019), effective September 1, 2019, added subsections (d) and (e) to Election Code § 255.004. A person commits an offense if, with intent to injure a candidate or influence the result of an election, the person creates a deep fake video and causes it to be published or distributed within 30 days of an election. "Deep fake video" means a video, created with the intent to deceive, that appears to depict a real person performing an action that did not occur in reality. An offense under the section is a Class A misdemeanor. SB 751, Sections 1 and 2; see also Election Code § 255.004(c) as quoted in introduced SB 1044

The 30-day window remains the law. HB 401 (89th Legislature, 2025) proposed striking the 30-day limitation so the offense would reach publication or distribution at any time, but the bill was left pending in the House Elections Committee after its April 3, 2025 hearing and never advanced. It should be described as a proposal, not as enacted law. HB 401 bill history, capitol.texas.gov

The constitutional history of the section concerns a different subsection. In Ex parte Stafford, the Dallas Court of Appeals held Election Code § 255.004(b), the "true source" provision punishing misrepresentation of a campaign communication's source, facially unconstitutional because the content-based restriction on political speech was not narrowly tailored to the State's compelling interests. Ex parte Stafford, 667 S.W.3d 517 (Tex. App. Dallas 2023) The Court of Criminal Appeals granted discretionary review and agreed, holding that subsection (b) does not pass constitutional muster. Ex parte Stafford, No. PD-0310-23 (Tex. Crim. App. 2024) Neither opinion addressed subsection (d), the deepfake provision, which was not at issue in the case.

The SCOPE Act: Platform Duties Toward Known Minors

The Securing Children Online through Parental Empowerment Act, HB 18 (88th Legislature, 2023), added Business and Commerce Code chapter 509, effective September 1, 2024 for the chapter's operative provisions. It is not an AI statute. Its duties fall on digital service providers, and they reach synthetic media only insofar as that media falls within the chapter's defined categories of harmful material or data practices.

A digital service provider may not enter into an agreement to create an account unless the person registers the person's age, and a person who registers as younger than 18 is a known minor. The provider may not let the user alter a registered age without a commercially reasonable review process, and a minor is also a known minor if a parent or guardian notifies the provider, successfully disputes the registered age, or otherwise acts under the chapter. HB 18, Sections 509.051

For a known minor, the provider must limit collection and use of personal identifying information to what is reasonably necessary for the service, and may not permit purchases or financial transactions, share or sell the minor's personal identifying information, collect precise geolocation data, or display targeted advertising to the minor, unless a verified parent provides otherwise. HB 18, Section 509.052 The provider must create parental tools letting a verified parent supervise the minor's use, including controlling privacy and account settings. HB 18, Section 509.054

The provider must also develop and implement a strategy to prevent a known minor's exposure to harmful material and content promoting, glorifying, or facilitating suicide, self-harm, eating disorders, substance abuse, stalking, bullying, harassment, or grooming, trafficking, child pornography, or other sexual exploitation or abuse, using specified measures including filtering technology, hash-sharing, and human monitoring reviews. Synthetic sexual material involving a minor falls within the sexual-exploitation category of that duty as written. HB 18, Section 509.053

Enforcement runs through the Attorney General. A violation of the chapter is a deceptive act or practice actionable under Subchapter E, Chapter 17, solely as an enforcement action by the consumer protection division of the Attorney General's office. HB 18, Section 509.151 The chapter otherwise provides no private right of action, except that the parent or guardian of a known minor affected by a violation may bring an action seeking only a declaratory judgment or an injunction; a court may not certify such an action as a class action. HB 18, Section 509.152

One duty is currently unenforceable. In CCIA v. Paxton, the Fifth Circuit affirmed on July 24, 2026, by a 2 to 1 vote, the preliminary injunction against the section 509.053 monitoring and filtering requirement, holding it likely preempted by Section 230 of the Communications Decency Act. The decision addressed only that preliminary relief, not the validity of the chapter as a whole, and Judge Ho partially dissented. Reuters, July 24, 2026

What Is Unsettled

Whether prosecutors will bring cases under the rewritten 21.165 and how courts will construe its consent and affirmative-defense provisions, which have no developed case law. Whether Election Code § 255.004(d) survives a First Amendment challenge if one is ever brought, given that Stafford invalidated only subsection (b). Whether the Fifth Circuit's preliminary view of section 509.053 holds on the merits, and how the remaining SCOPE Act duties apply to services whose content is substantially AI-generated.

Compliance Checklist

For anyone producing, distributing, or hosting synthetic media in Texas:

• Confirm written consent before producing or distributing sexually explicit synthetic media depicting a real person. Under § 21.165, only a knowingly and voluntarily signed plain-language written agreement describing the media counts, and the signed agreement must identify any audiovisual work incorporating the media.
• Treat a "this is fake" label or disclaimer as legally worthless under § 21.165. The statute expressly says it is not a defense.
• Calendar the felony triggers: a prior conviction under § 21.165 or a depicted person younger than 18 makes the production or distribution offense a third-degree felony, and threats involving a minor raise the threat offense to a Class A misdemeanor.
• If you operate a conduit service, confirm your role is solely technical, automatic, or intermediate before relying on the (c-4) defense. If you provide a publicly accessible AI application, put the statutory prohibition in terms users must acknowledge before access and implement the technological prevention steps the statute lists before relying on the (c-5) defense. Neither defense turns on whether you removed material after notice.
• For election content, assume any synthetic video depicting a real person, created with intent to deceive and published or distributed within 30 days of an election with intent to injure a candidate or influence the result, is a Class A misdemeanor under Election Code § 255.004(d). Do not rely on a repeal of the 30-day window; HB 401 did not pass.
• If your service may be used by minors, build the SCOPE Act duties around age registration, data minimization, parental tools, and the harmful-material strategy, and track the CCIA v. Paxton litigation before treating the 509.053 filtering requirement as enforceable.
• Remember the separate civil track: Civil Practice and Remedies Code chapter 98B claims, the 72-hour site removal duty, social-media complaint systems, and the 48-hour duty under the federal TAKE IT DOWN Act are covered in the companion guide and run alongside these criminal and platform provisions.