What the Act Means by “Digital Forgery” and “Consent”
The TAKE IT DOWN Act turns words that sound familiar—consent, identity, intimate image, and deepfake—into legal categories with different consequences. This chapter focuses on those categories so readers can examine a reported image in the right order: identify the person, identify the kind of depiction, identify what authorization existed, and then notice the contextual limits that may matter. That discipline is useful because calling something “AI-generated” does not, by itself, answer what the Act covers or what anyone must prove.
The definitions also explain why the 48-hour platform rule cannot be separated from classification. A platform must be able to locate the depiction and decide whether a request is valid, but the criminal provisions ask more specific questions about the material and the publisher's conduct. The same vocabulary supports both tracks, yet the tracks do not use every term in exactly the same way.
The legal object is a person-linked intimate depiction
The Act does not define a violation around technology alone. Its starting point is an “intimate visual depiction,” a term the statute adopts from an existing federal definition. The depiction must also be connected to an identifiable individual. Under the Act, that person may appear in whole or in part, but their face, likeness, or another distinguishing characteristic must be displayed in connection with the intimate visual depiction.
That person-link is important. A fabricated body with no identifying connection to anyone raises a different question from an image that uses a recognizable face, unique birthmark, or other feature to attach intimate content to a real individual. The Act's concern is not merely that software generated sexual imagery. It is that intimate material is connected to someone whom the law can identify as the depicted individual.
The law's definition of consent is similarly specific. Consent means affirmative, conscious, and voluntary authorization made free from force, fraud, duress, misrepresentation, or coercion. It is not a presumption based on silence, a past relationship, or the fact that someone once agreed to something related. The definition matters most when a publisher tries to treat an earlier permission as a general license for later disclosure.
The statute makes that distinction explicit in two rules. Consent to the creation of an intimate visual depiction does not establish consent to publish it. And the fact that an individual disclosed the depiction to another person does not establish consent for the person accused of violating the law to publish it. In practical terms, permission can be limited by purpose, audience, and act. Creating an image and distributing it are separate events; sending it privately to one person and publishing it through a service are also separate events.
This is why “nonconsensual” should not be treated as a single, self-explanatory fact. A careful account asks: What exactly did the individual authorize? To whom was it given? Did the authorization cover publication, or only creation or private disclosure? Was the authorization free from the forms of pressure or deception named in the statute? Those questions do not require a reader to decide a criminal case, but they determine which facts are relevant to the Act.
The Act also defines a minor as an individual under 18. That status matters because the criminal provisions later use a different structure for depictions involving minors. It is not safe to learn the adult rules and assume they simply apply to everyone. The age category is part of classification before it is part of consequence.
Identifiability does not necessarily mean that an entire face appears. The statute expressly allows the individual to appear in whole or in part, so a partial depiction can still raise the statutory question if a face, likeness, birthmark, or other recognizable feature connects it to that person. The reverse is also important: a report may name someone, but the legal analysis still has to ask whether the depiction itself displays the identifying connection. A name supplied outside the image is not the same thing as a characteristic displayed in connection with it.
Consent should be analyzed with the same care. The statutory definition does not ask whether the publisher and the depicted person once knew each other or whether some related image was previously shared. It asks whether there was affirmative, conscious, voluntary authorization and whether that authorization was free from the forms of pressure and deception named in the law. The two rules about creation and disclosure then prevent a publisher from converting a limited permission into a broader one by assumption.
This does not make every consent dispute easy. The retained sources do not prescribe a particular document, message, or authentication method that proves consent, and they do not explain how a platform must resolve every conflicting account. What the statute does provide is a disciplined set of questions. A platform reviewing a notice, a prosecutor evaluating a possible offense, and a reader assessing a public claim may need different levels of proof, but all should avoid treating a relationship, a prior image, or a technology label as a substitute for the relevant authorization and identity facts.
A digital forgery is narrower than everyday “AI image”
The Act defines a digital forgery as an intimate visual depiction of an identifiable individual created through software, machine learning, artificial intelligence, or another computer-generated or technological means. The definition includes adapting, modifying, manipulating, or altering an authentic depiction. But the technology requirement is only part of the test. When viewed as a whole by a reasonable person, the result must be indistinguishable from an authentic visual depiction of the individual.
That final condition does important work. In ordinary conversation, people may call any synthetic or edited image a deepfake. The statutory category is narrower. A technology-created image can be relevant to the subject without automatically satisfying the Act's definition of “digital forgery.” If the image is obviously fictional, visibly implausible, or otherwise not indistinguishable as a whole, its status under this particular definition becomes a threshold question. The available sources do not say that every such image falls into another TIDA category, another law, or no legal protection at all. They support caution, not a shortcut.
The reasonable-person test also focuses on the depiction as a whole. That does not mean every viewer must be fooled or that a technical expert's ability to detect manipulation decides the issue. It means the statutory definition asks how the resulting depiction would appear to a reasonable person when considered in its entirety. A technical claim that software was used is therefore not enough by itself; the character of the finished depiction matters too.
The distinction between authentic material and digital forgery becomes sharper in the adult criminal provisions. For an adult authentic intimate visual depiction, the statute asks whether the depiction was obtained or created in circumstances in which the publisher knew or reasonably should have known the individual had a reasonable expectation of privacy. For an adult digital forgery, the statute instead expressly asks whether the material was published without the identifiable individual's consent. Both branches contain additional requirements: what is depicted must not have been voluntarily exposed by the individual in a public or commercial setting, it must not be a matter of public concern, and publication must be intended to cause harm or cause harm, including psychological, financial, or reputational harm.
These are not interchangeable ways of saying “the person did not like the image.” The authentic branch focuses on the circumstances in which the material was obtained or created and the person's expectation of privacy. The digital-forgery branch puts publication without consent at the center. That difference does not mean the authentic branch ignores consent or the digital branch ignores context; it means the statutory gates are arranged differently.
The public and commercial exposure condition creates another boundary. An adult's intimate depiction is not placed in the same position when what it shows was voluntarily exposed in a public or commercial setting. The public-concern condition creates a related limit for material that may concern matters of legitimate interest to the community. The Congressional Research Service says this language appears connected to a First Amendment concept under which speech about political, social, or other matters of public concern generally receives greater protection than speech about purely private matters. The source identifies the legal context; it does not decide how a court would apply the condition to a particular image.
The same caution applies to the platform provisions. The criminal subsections separately name authentic intimate visual depictions and digital forgeries. The notice-and-removal provisions, however, speak of an intimate visual depiction published without the individual's consent rather than repeating the separate criminal label. The CRS analysis identifies a real interpretive question: perhaps the terms mark distinct materials throughout the Act, or perhaps a digital forgery is one kind of intimate visual depiction for the platform and civil-remedy provisions. The definitions suggest overlap, but the supplied sources do not resolve every borderline case.
Classification is a gate, not a verdict
The diagram's order is deliberate. A reader cannot responsibly discuss consent without knowing whose consent is relevant. They cannot apply the digital-forgery definition without asking whether the material is an intimate visual depiction, whether the individual is identifiable, whether technology created or altered it, and whether the finished result meets the reasonable-person condition. They cannot apply the adult criminal branches without carrying forward the public-exposure and public-concern limits.
But passing through a classification gate does not establish criminal liability. The adult offenses also require knowing publication through an interactive computer service, the applicable harm or intent condition, and other statutory facts. The words “publish” and “knowingly” are not defined in the Act or the amended statute, according to the CRS. The definitions in this chapter tell a reader what kind of object and authorization to examine; they do not supply every fact or resolve every legal interpretation.
The categories also protect against two opposite mistakes. The first is underinclusion: treating a digital forgery as outside the law simply because no authentic photograph was taken. The Act expressly includes technological creation and alteration in its digital-forgery definition. The second is overinclusion: treating every edited or AI-assisted intimate image as a qualifying digital forgery, or treating every accusation of nonconsent as proof of a criminal offense. The statutory conditions are more specific than either assumption.
For the 48-hour removal process, this means a request can be urgent without being rhetorically inflated. The individual needs to identify the depiction and state a good-faith belief that it was published without consent. The platform must then apply its statutory process, not wait for a criminal conviction. At the same time, readers should not infer that a removal request proves that the material meets every criminal definition. Platform containment and criminal adjudication use related facts for different institutional purposes.
The most durable distinction is therefore simple: technology describes how an image may have been made, while the Act asks what the image depicts, whom it identifies, what permission existed, and how the material was published. Those questions carry forward into the criminal rules. There, the difference between authentic depictions and digital forgeries becomes a difference in what must be proved—not merely a difference in vocabulary.