Let’s start with the basics \u2013 what are deepfakes exactly?<\/h2>\n
Given the origins of the name, you might think that this is a new thing, and it kind of is but is older than you might think. The idea of photo-manipulation goes all the way back to the 19th century when photography was invented<\/strong>.<\/p>\n But let\u2019s be real, that\u2019s not what\u2019s interesting about this phenomenon.<\/p>\n A deepfake is synthetic or altered media made with AI to imitate a person’s appearance or voice. A face swap is one example; a fabricated voice recording is another. The convincing part is that the result can look or sound like evidence of something the person never did.<\/p>\n Not every edited image is a deepfake. Ordinary retouching, cutting a video out of context and changing playback speed can mislead people without AI. The label describes a technique, not a guarantee that the result is either believable or harmless.<\/p>\n In adult content, the crucial question is consent. A person agreeing to a photograph, acting role or public social-media profile has not thereby agreed to be depicted in sexual material. A recognizable face does not come with permission attached.<\/p>\n First, the honest answer: you cannot reliably settle this just by staring at a face.<\/strong> Visual oddities are reasons to check further, not proof. A polished fake may have none of them, while a genuine clip may look strange after compression or editing.<\/p>\n NIST’s review of synthetic-content detection<\/a> describes limitations in both human judgment and automated detectors. A detector’s performance can depend on the generators and material it was tested on. Treat a confident percentage as a clue, not a verdict.<\/p>\n A mouth that drifts around a face or speech that does not match lip movements deserves a closer look. But dubbing, an audio delay or a poor export can also cause a mismatch. Check the original upload if you can find it; a cropped repost is a lousy foundation for certainty.<\/p>\n Odd reflections, gaze or blinking can raise questions. They do not give you a universal deepfake test. Nor does normal blinking prove that a clip is genuine. Avoid turning one visual quirk into a public accusation about the person shown.<\/p>\n Look for inconsistent edges around the face, lighting that does not fit the scene, or details that change between frames. Then ask what else could explain them. Filters, motion blur and low resolution can hide useful detail or create misleading artifacts.<\/p>\n A jump or change in speed can come from ordinary editing. A real recording can also be given a false caption or stripped of context. The practical question is broader than ‘Was AI used?’: does the source actually support the claim being made?<\/p>\n Fame does not establish authenticity, and it does not establish consent. A sexual image can be fabricated, stolen or shared without permission. Do not assume it must be fake because the person is famous, or legitimate because the page calls it a leak.<\/p>\n Find the earliest available publication, check who posted it and look for independent reporting that actually verifies it. Ten sites repeating the same anonymous claim are still one unverified claim. If you cannot establish the context, say you don’t know. You do not need to circulate an intimate image to ask whether a claim is credible.<\/p>\n At a high level, these systems learn patterns in images, video or audio and use them to generate or alter media. A face-swap system changes who appears in a scene; other systems synthesize images or imitate voices. Explaining the technology does not require handing out tools for fabricating intimate images of real people.<\/p>\n One widely repeated statistic needs its date attached. Deeptrace’s September 2019 report<\/a> identified 14,678 online deepfake videos, with 96% classified as pornographic. Its comparison was with 7,964 videos measured in December 2018, not January 2019.<\/p>\n That was a historical study of the material the researchers identified, not a count of everything online today.<\/strong> It does not establish the current share of AI images, private messages or all synthetic media. The finding matters; stripping away its date and scope makes it less useful.<\/p>\n Let’s be blunt: ‘it’s fake’ does not mean ‘nobody gets hurt.’ And ‘they’re famous’ is not permission. The person being impersonated may have to deal with harassment or explain a fabricated image to people who should never have treated it as real in the first place.<\/p>\n Law depends on the place, the conduct and the facts. Creating an image, sharing it and threatening someone with it can raise different issues. A copyright argument does not settle consent or every other legal question. Do not treat a parody label as a free pass.<\/p>\n England and Wales:<\/strong> the old version of this article suggested that significantly modified intimate images could simply be created and shared in England. That advice is outdated. The Crown Prosecution Service’s current guidance<\/a> covers sharing or threatening to share intimate images, and offences of creating or requesting intimate deepfakes of adults without consent or a reasonable belief in consent. The creation and request offences came into force on February 6, 2026.<\/p>\n United States:<\/strong> the FTC’s Take It Down Act guidance<\/a> says covered platforms must provide a reporting process and remove reported nonconsensual intimate imagery and known identical copies within 48 hours of a valid request. The requirement includes digital forgeries and took effect on May 19, 2026. It is not a promise that every copy will vanish from the internet.<\/p>\n Those are specific examples, not a worldwide legal ruling. For a real case, use current local advice. My editorial line is simpler: a person’s face is not a public resource for sexual impersonation.<\/p>\n You should not have to become a video-forensics expert before asking for help. Keep a private record of the page addresses, account names, dates and report numbers. Avoid reposting the material to prove that it exists; that can give it another audience.<\/p>\n A friend helping you can organize links and reports without forwarding the image around. If threats are involved, seek local victim support or legal help; do not let the person threatening you dictate the entire conversation.<\/p>\n My regular readers know I have plenty of opinions. On this one, the standard is straightforward: understand the technology, question the source and respect the person whose face is involved. A convincing image is not evidence of consent.<\/p>\n For the wider industry discussion, read Deepfake Porn Explosion: How it’s Spiraling Out of Control<\/a>. This explainer covers how to think about a suspect image; that article looks at the wider problem and platform responsibility.<\/p>\n","protected":false},"excerpt":{"rendered":" The words ‘deep learning’ and ‘fake’ tell you where the name comes from. They don’t tell you whether a clip is real, whether the person agreed to it, or what happens when a convincing fake gets loose online. That’s where this deep dive comes in. Let’s look at how deepfakes work, the clues worth checking, … Continue reading Let\u2019s do a deep dive on deepfakes<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1124],"tags":[564,1238,11],"class_list":["post-18860","post","type-post","status-publish","format-standard","hentry","category-porn-tips","tag-deepfakes","tag-definitive-guide","tag-porn"],"acf":[],"_links":{"self":[{"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/posts\/18860","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/comments?post=18860"}],"version-history":[{"count":7,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/posts\/18860\/revisions"}],"predecessor-version":[{"id":31620,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/posts\/18860\/revisions\/31620"}],"wp:attachment":[{"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/media?parent=18860"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/categories?post=18860"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/theporndude.com\/blog\/wp-json\/wp\/v2\/tags?post=18860"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}How to check a suspected deepfake without fooling yourself<\/h2>\n
![\"\"](\"https:\/\/theporndude.com\/blog\/wp-content\/uploads\/webp\/content\/2021\/01\/What-is-Legit-and-What-is-a-Deepfake.webp "\\"\\"")
<\/picture><\/p>\nFunny moving lips<\/h3>\n
The eyes look ‘off’<\/h3>\n
Other unnatural facial and body cues<\/h3>\n
There are sudden changes in movement<\/h3>\n
A celebrity’s face is not proof of anything<\/h3>\n
Check the source before you share<\/h3>\n
What the software does, and what the numbers really say<\/h2>\n
![\"\"](\"https:\/\/theporndude.com\/blog\/wp-content\/uploads\/webp\/content\/2021\/01\/This-is-the-sort-of-software.webp "\\"\\"")
<\/picture><\/p>\nIs this legal, and where does consent fit?<\/h2>\n
What to do if a deepfake targets you<\/h2>\n
\n
ThePornDude\u2019s final words about deepfakes<\/span><\/h2>\n