Skip to content
The XXXPress
The XXXPress
Legal & PolicyAnalysis · Includes sourced legal commentary. This is not legal advice; consult qualified counsel on your own obligations.

The Consent Standard: What the AI Era Is Asking of Adult Companies

Europe's transparency rules and the first US deepfake prosecutions are usually reported as separate stories. Read together, they describe a single new operating requirement: consent you can prove, and synthetic content you can label.

By The Industry Business Desk··8 min read·Updated 11/19/2026
The XXXPress
Advertisement

Your ad here

Advertise with us

Between the EU AI Act's transparency duties and the first prosecutions under America's TAKE IT DOWN Act, the industry's next competitive advantage is paperwork that survives scrutiny: written likeness rights, visible labeling, and real human review.

In May 2026, federal prosecutors in Brooklyn unsealed charges against two men accused of publishing thousands of AI-generated images and videos depicting real people nude or engaged in sexual acts. The victims named in the complaints included actresses, singers and political figures, along with people who are not public figures at all. In October, an industry attorney's analysis of Europe's AI Act set out a list of obligations that adult platforms are already living with: disclose synthetic content, label deepfakes, stop inferring sexuality from biometrics, and rewrite performer releases that were never drafted to cover a world of digital replicas.

Reported separately, those two items read as enforcement news and legal commentary. Read together, they are a specification. The industry is being asked to prove consent rather than assume it, to mark synthetic work rather than let it blend in, and to keep a human being accountable for decisions a model makes.

That is a higher bar than most companies currently meet, and it is also an opportunity to lead rather than be dragged.

The rules, in two paragraphs

The European Union's AI Act is now largely in force, and it follows providers and deployers outside Europe when their systems are used on EU users. Operators using third-party AI are generally deployers; companies that develop, brand or materially modify systems can become providers, with heavier obligations. The transparency duties are practical rather than abstract: users interacting with an AI chatbot, companion or synthetic performer must be told they are dealing with AI at the start of the interaction, and deepfake content must be disclosed by the viewer's first exposure. A machine-readable watermark alone is not treated as sufficient, and creative or artistic framing does not create an exemption.

The prohibitions are where the sharper change sits. Per the analysis published in October, the 2026 omnibus amendments forbid AI systems intended to generate or manipulate nonconsensual sexually explicit or intimate material, including nudification tools, and systems intended to generate child sexual abuse material, including synthetic material, with application from 2 December 2026. The Act also forbids using biometric data to infer protected characteristics such as a person's sex life or sexual orientation. Penalties for prohibited practices reach EUR 35 million or 7% of worldwide annual turnover; other violations, including transparency failures, reach EUR 15 million or 3%.

In the United States, the TAKE IT DOWN Act, enacted on 28 April 2025, criminalizes the nonconsensual publication of intimate images including AI deepfakes. The Eastern District of New York prosecutions are the clearest signal yet that it will be used: one defendant is alleged to have published at least 360 albums of deepfake material depicting roughly 90 women, viewed millions of times; the other is alleged to have published about 113 albums depicting roughly 50 identifiable women, viewed nearly a million times. Each faces up to two years' imprisonment if convicted. The charges are allegations, and both defendants are presumed innocent.

The consent gap inside a standard release

Most performer releases in circulation were written for a world where a scene was shot once, in a room, with a person present. They generally do not say whether the footage may be used to train a model, whether a face may be replaced, whether a voice may be cloned, or what happens when a performer wants an authorisation withdrawn.

That is the gap the AI Act analysis points at when it advises operators to address digital replicas, voice cloning, model training and face replacement expressly and separately rather than inside broad boilerplate. Consent that a performer did not specifically give for synthetic reuse is not consent a company should want to rely on.

The operational fix is unglamorous. Separate permissions for separate uses. Named duration. Stated compensation. A documented revocation path with a defined response time. Records that show which version of which release applied to which shoot, stored somewhere a lawyer can find in an afternoon. Companies that do this well end up with something better than compliance: a reputation among talent that their likeness is handled like an asset rather than a byproduct.

Labeling is a product decision

The common mistake is treating labeling as a checkbox in an upload flow. The rules are explicit that a buried term-of-service line will not do.

There is a design choice here, and it is a marketing one. A synthetic performer that is openly synthetic, labeled in the player and in the profile, can build a following on its own terms. The alternative, an undisclosed replica of a real person, is the version that produces criminal charges, platform removals and headlines that damage every company in the category.

The EU analysis notes that even where a synthetic performer does not resemble any real person and therefore may not meet the legal definition of a deepfake, voluntary labeling remains sensible risk management. That is the tell. The companies likely to navigate this well are the ones labeling more than they are required to, because they have decided trust is cheaper to maintain than to rebuild.

Human oversight is an operating model, not a slogan

"Human in the loop" is easy to write in a policy and hard to staff. In practice it means a named reviewer for consequential automated decisions affecting performers, a documented appeal route, and logs showing that a person actually looked before an account, a payout or a listing was affected.

It also means being honest about where automation should not decide. The AI Act analysis draws a line that matters commercially: systems used to recruit, evaluate or allocate work to people can fall into the high-risk category, with the principal requirements for those systems scheduled from 2 December 2027. A performer-management workflow that ranks or deactivates creators on automated criteria is a different risk conversation from a moderation filter that flags an upload.

Detection itself is not automatically high-risk, but accuracy, bias and foreseeable harm still need evaluating.

AdvertisementThe XXXPress

Your ad here

Reach the executives, lawyers and operators shaping the industry.

Advertise with us

The privacy blind spot

The prohibition on inferring sexual orientation or sex life from biometric data is the provision most likely to catch an operator by surprise, because the technology it forbids is commercially available and often sold as personalisation. Inferring preferences from a face, a voice or a body shape is now the wrong side of a line.

Recommendation based on viewing activity is different, though behaviour patterns can themselves reveal special-category data under GDPR.

Age assurance carries the same tension: facial estimation is not banned but is biometric categorisation, so it should be disclosed, minimal, and never the excuse to build a permanent identity-linked database.

Where leadership actually shows

A handful of behaviours separate companies that will look credible in 2027 from those explaining themselves.

Published standards. A public page describing how synthetic content is labeled, what consent is required before a likeness is used, and how a performer requests removal.

Performer-facing controls. A dashboard where a creator can see and revoke authorisations, rather than an email address that may or may not be monitored.

Vendor diligence with teeth. Data retention, training use, processing locations, deletion, incident notification and audit rights written into AI supplier contracts, not assumed.

Training. The AI-literacy obligation is already in effect and should map to what each team touches.

Records. Regulators and plaintiffs both work from documents, and an AI inventory plus a release history is worth more than any assurance.

None of this requires slowing down. AI is not prohibited in this business; secrecy, nonconsensual generation and uncontrolled biometric analysis are what is becoming untenable, and companies that build openly will find the technology easier to use, not harder.

Sourced commentary, not legal advice

The legal points in this story are drawn from published analysis by counsel and from the text of the enforcement documents linked below, and they are reported as commentary rather than conclusions. Obligations differ by jurisdiction, by the operator's role in the AI supply chain, and by facts that only your own adviser can assess. Companies should treat this as a map of the questions to bring to qualified counsel, not as an answer.

Read the original reporting

  • XBIZ, "What Adult Platforms Should Know About the EU AI Act" (14 October 2026)
  • U.S. Attorney's Office, Eastern District of New York, "Two Individuals Arrested for Publishing AI Deepfake Pornography in Violation of TAKE IT DOWN Act" (20 May 2026)
  • Federal Trade Commission, Take It Down reporting portal

Explore the desk

Attribution: Facts and legal analysis are drawn from a published industry legal analysis of the EU AI Act (October 2026) and a U.S. Department of Justice press release (May 2026). Charges described are allegations; defendants are presumed innocent.

About this byline

The Industry Business Desk

The Industry Business Desk covers the operating side of the business: contracts, compliance, risk, payments and the management decisions that determine which companies survive regulatory change.

Reported and written by XXXPress editorial staff from the public documents linked in this story. Legal points are attributed to published analysis by counsel and are presented as sourced commentary, not legal advice.

Fact-checked by the desk before publication

Sources and further reading

Original reporting by The XXXPress. The links below are attributed to their publishers for verification and context; we do not host or republish their articles.

Read the original source · XBIZ / U.S. Department of Justice

#EU AI Act #TAKE IT DOWN Act #consent #digital replicas #deepfakes #likeness rights #compliance

Read next

The XXXPress Top 100 Performers of 2026

Performers & Creators · 2 min read

Relevant jobs

All jobs →

Nearby events

  • No dates listed yet.

Performers and studios

Browse directory →

The XXXPress Radio

Listen while you work

Continuous music, plus the club and appearance calendar for the week ahead.

Open the player

The Daily Brief

Industry intelligence, every morning.

Business, technology and regulatory news in one concise email.

Advertise

Reach the people who run the industry.

Listings from $19.99 per month. Always clearly labeled.

View packages

More from The XXXPress

View all →