Synthetic adult video still needs a business that can distribute it
A generated clip leaves payment access and consent unresolved. Current platform obligations make those operating costs part of the product.

Generating a convincing clip answers one production question. Selling access to it creates a different set of obligations. A business needs a distributor willing to carry the content and a payment relationship that covers the actual product. A person affected by an unauthorized depiction also needs a working route to have it removed.
These constraints help explain why an impressive model demonstration says so little about a sustainable adult-media service. The public evidence does not support a sweeping claim that synthetic adult video is absent, nor does a visible example establish a profitable market. A useful assessment starts with the operating conditions a particular business can document.
The January 2026 debate often treated those conditions as something that would be solved after video quality improved. By November, one relevant US platform deadline had already passed. The Federal Trade Commission began enforcing Section 3 of the TAKE IT DOWN Act on May 19, 2026. Covered platforms must provide a removal process for nonconsensual intimate imagery and remove it, including known identical copies, within 48 hours of a valid request.
Payment access depends on the actual service
A legal product can fall outside a payment provider's supported categories. Stripe's current adult-content guidance says businesses selling or offering adult content or services are unsupported. A general account approval for another business activity cannot establish that a new adult-media offering is supported.
Card networks and payment processors also occupy different positions. Visa's network-integrity policy allows legal transactions subject to its rules, with additional requirements for acquirers serving adult-content merchants. Visa explicitly applies its prohibitions on nonconsensual sexual content and child sexual abuse material to computer-generated content too. That policy differs from a processor declining the category altogether.
For a business proposal, the distinction changes what evidence is needed. A founder should be able to identify the provider that has agreed to process the described service. The description needs to match what customers will receive. A hypothetical provider relationship contributes nothing to the cash-flow calculation until that provider accepts the activity.
This is not an invitation to disguise the product. Concealing its nature leaves the business dependent on a relationship formed under inaccurate information. A service that loses payment access can have a functioning video pipeline and still be unable to collect revenue or manage existing customers properly.
Consent has to survive a complaint
A fully fictional character and a synthetic depiction of a recognizable person create different factual questions. Applying a blanket label such as AI-generated does not resolve whether someone's likeness was used without permission. An operator needs a way to investigate that claim against the particular content being reported.
The FTC's May enforcement announcement puts a time limit around one part of that work for covered US platforms. The 48-hour requirement follows a valid removal request. It should not be paraphrased as a universal deadline applying to every service worldwide or as a complete summary of the law's criminal provisions. Coverage and other legal duties depend on the facts of the service.
Operationally, someone has to receive the request and act on it. If a small team only reviews complaints during a weekly meeting, its schedule cannot support a two-day removal obligation. The removal process must also reach the places where the service publishes the material. Deleting a database entry may leave a separately hosted file available.
A useful internal exercise follows one test asset through publication and withdrawal. It asks who has authority to disable access and how the team can confirm the result. That exercise can expose a missing owner without creating harmful content or using a real person's likeness. The goal is to test the service's response process with harmless material.
Count the work after generation
A low advertised generation price covers a limited part of production. Suppose a business pays for several attempts before one clip meets its quality requirements. The accepted clip carries the cost of those failed attempts. Reviewing it also consumes time, even when the final file needs no editing.
The same accounting applies to content administration. Staff time spent assessing a complaint belongs in operating cost. A payment provider's requirements can add work before a service starts accepting customers. Those expenses may be manageable, but excluding them makes the comparison with conventional production misleading.
A credible cost estimate can use an actual trial with non-explicit test footage. Record how many attempts produce an acceptable result and how much review time each attempt needs. Keep the quality threshold fixed during the trial. Lowering it to improve the success rate changes the product being costed.
Demand then needs its own evidence. Revenue from a conversational companion service does not measure willingness to pay for a separate video product. A founder can test the intended offer through lawful research with consenting adults, without treating another category's sales as a forecast.
Before scaling a generation pipeline, resolve the distribution relationship for the exact service. Then cost the review and removal work under that arrangement. Those results will tell a team how much production it can support before another model improvement becomes commercially relevant.