When AI-Generated Images Become Illegal

The first mistake is assuming an AI-generated image is legally safer because it is "not real."

That is sometimes true in ordinary creative work. It is not a reliable rule. A synthetic image can still be illegal if it depicts child sexual abuse, uses a real person in a nonconsensual intimate deepfake, is used for threats or extortion, falls within an obscenity statute, impersonates someone in a fraud scheme, or creates election or consumer deception covered by another law.

A recent Lubbock report is a useful example. EverythingLubbock reported that a local defendant was indicted for creating and distributing bestiality material. An indictment is only an accusation. But the story raises the right AI-governance question: when does generated media stop being "fake content" and become prohibited material?

The Lubbock Case: Bestiality, Obscenity, and Material

The public report describes an indictment involving bestiality material. Without repeating the defendant's identity, the Texas law backdrop matters.

Texas Penal Code Section 21.09 criminalizes bestiality conduct and related conduct, including organizing, promoting, conducting, observing, aiding, or permitting specified acts. Separately, Texas obscenity law defines "obscene" material to include, among other requirements, patently offensive representations or descriptions of sexual bestiality when the material also appeals to prurient interest and lacks serious literary, artistic, political, or scientific value.

That distinction matters for AI images. If a case involves real conduct with an animal, the bestiality statute may be the central issue. If the case involves images or files, obscenity and distribution concepts may become important. If minors or child-like depictions are involved, a different set of child-exploitation or obscene-child-depiction laws may apply. The same generated file can therefore raise more than one legal question.

AI Is Usually Not the Legal Element

Most laws do not ask, "Was this made with a diffusion model?" They ask more concrete questions.

• ●What does the image depict? Child sexual abuse, sexual bestiality, nudity, violence, private anatomy, election conduct, regulated products, identity documents, or something else.
• ●Who does it appear to depict? A real adult, a real minor, an identifiable person, a public figure, a fictional person, or no recognizable person.
• ●How realistic is it? Some statutes turn on whether the synthetic image is indistinguishable from a real person or appears authentic to a reasonable person.
• ●Was there consent? Consent is central for intimate deepfakes and image-based abuse laws.
• ●What did the creator or distributor do with it? Possession, production, distribution, threats, sale, business promotion, publication, and harassment can trigger different rules.
• ●Is there another unlawful purpose? Fraud, extortion, stalking, identity theft, election deception, or consumer deception can make the image part of a broader illegal act.

Minors Are the Brightest Red Line

The clearest category is child sexual abuse material and obscene depictions of minors. Federal law already reaches more than traditional photographs.

The Justice Department explains that federal child-pornography law covers visual depictions including photographs, videos, digital images, computer-generated images indistinguishable from an actual minor, and images adapted or modified to appear to depict an identifiable actual minor. Another federal statute, 18 U.S.C. Section 1466A, covers obscene visual representations of the sexual abuse of children and does not require prosecutors to prove that the depicted minor actually exists.

That is why recent federal cases have included AI-generated images. DOJ has announced convictions and sentences where defendants used AI tools to produce or possess obscene child-abuse depictions, sometimes alongside real CSAM and sometimes using faces or likenesses of children from the community. The practical rule is blunt: do not treat synthetic minor sexual content as a loophole.

Texas Added a Child-Like Visual Material Rule

Texas has also moved directly into synthetic-child territory. Penal Code Section 43.235, added in 2025, covers possession, access with intent to view, promotion, and production of obscene visual material that appears to depict a child younger than 18 engaged in specified sexual conduct. The statute expressly reaches material regardless of whether the depiction is an actual child, a cartoon or animation, or an image created with an AI application or other computer software.

Texas also increased punishment under its general obscenity statute when the obscene material visually depicts covered sexual activities involving a real child, an image virtually indistinguishable from a child, or an identifiable child whose image was created, adapted, or modified.

For companies, schools, platforms, and internal teams, the governance lesson is simple: moderation rules should not wait to determine whether an image is "real" before escalating apparent minor sexual content.

Nonconsensual Intimate Deepfakes Are Now a Core Enforcement Target

A second major category is nonconsensual intimate imagery. These laws are aimed at "nudify" apps, sexual deepfakes, revenge-porn-style distribution, and threats to publish fake intimate images.

The federal TAKE IT DOWN Act, enacted in 2025, created criminal prohibitions around publishing or threatening to publish certain nonconsensual intimate images, including AI-created "digital forgeries." It also requires covered online services to remove qualifying material after a valid notice process. DOJ has already announced a guilty plea that included a first-in-the-nation conviction under the new federal digital-forgery provision.

Texas has a separate deepfake sexual-media statute. Penal Code Section 21.165 prohibits, without effective consent, knowingly producing or electronically distributing deepfake media that appears to depict a real person with intimate parts or engaging in sexual conduct they did not engage in. It also covers certain threats to produce or distribute such media. The statute says a disclaimer or label that the media is fake is not a defense.

States Are Building Their Own Deepfake Rules

The United States is becoming a patchwork. Texas is one model. California is another.

California signed SB 926 in 2024 to create a new crime targeting AI-generated sexually explicit deepfake content. California also has civil remedies for people depicted in nonconsensual sexually explicit synthetic media. In 2026, California officials also publicly pressed AI companies over nonconsensual intimate images and CSAM generated through consumer AI systems.

The trend is broader than any one state: lawmakers are treating sexual deepfakes as image-based abuse, not as ordinary parody or harmless fabrication. The practical compliance issue is not only user behavior. It is whether a product gives users the tools to create prohibited images, whether policies ban that use, whether filters and reporting tools work, and whether the company responds quickly when abuse is reported.

Other Countries Are Taking Different Approaches

Jurisdictions are not converging on one single rule.

The United Kingdom has moved toward criminalizing creation and requesting of nonconsensual intimate deepfakes, while sharing or threatening to share intimate images, including deepfakes, is already treated as an offense under its intimate-image framework. South Korea has gone further in some respects: after a wave of deepfake sex-crime scandals, legal revisions approved in 2024 criminalized possession, purchase, download, and viewing of certain deepfake sexual content, with higher penalties for production and distribution.

The European Union is taking a different layer of action through the AI Act. Article 50 focuses on transparency: deployers of AI systems that generate or manipulate image, audio, or video content constituting a deepfake must disclose that the content was artificially generated or manipulated, subject to exceptions. That does not replace criminal law. It adds a labeling and disclosure obligation on top of member-state laws and platform duties.

When AI Images Can Become Illegal

A practical issue-spotting framework is more useful than asking whether AI art is legal in the abstract.

What Businesses Should Put in Their AI Image Policy

Most organizations do not need a criminal-law memo for every AI image. They do need rules that keep ordinary teams away from obvious danger zones.

• ●Ban apparent minor sexual content outright. The review process should escalate immediately without debating whether the image is synthetic.
• ●Ban nonconsensual intimate or sexual likeness use. Do not generate, edit, publish, or test sexualized images of real people without documented consent and a legitimate use case.
• ●Do not permit bestiality, extreme sexual, or obscene content in business workflows. Even where First Amendment questions exist, this is not a sensible product, marketing, training, or internal-use risk.
• ●Block threats and coercive use. Policies should cover fake images used for blackmail, harassment, humiliation, workplace retaliation, or customer intimidation.
• ●Label synthetic media when it could mislead. For public-facing images, especially realistic people, products, events, or political content, disclose AI generation or manipulation where required or prudent.
• ●Keep an escalation path. Employees need somewhere to send questionable outputs, user reports, or vendor incidents without preserving or resharing harmful material unnecessarily.
• ●Review vendors. Ask image-generation vendors what they block, how they detect minors and real-person sexual deepfakes, what reporting tools exist, and how abuse logs are handled.

The Bottom Line

AI changes how quickly images can be made. It does not erase the law that applies to what those images depict or how people use them.

The Lubbock indictment is a useful reminder because it sits at the uncomfortable boundary between old categories and new tools: bestiality, obscenity, digital files, and possibly generated media. Other cases involving AI-generated child abuse imagery and sexual deepfakes show the same pattern. Prosecutors and lawmakers are not waiting for a perfect synthetic-media code. They are using existing child-exploitation, obscenity, harassment, extortion, privacy, and deepfake statutes, while adding targeted AI provisions where the old rules feel incomplete.

For governance purposes, the safest rule is direct: if an image appears to depict a minor sexually, sexualizes a real person without consent, threatens or coerces someone, promotes obscene material, impersonates someone for gain, or misleads the public in a regulated context, the fact that AI generated it is not a shield. It may be the reason the risk scaled so fast.

Related Regulations

Sources & References