This CSS code defines a custom font family called “Guardian Headline Full” with multiple font weights and styles. It includes light, regular, medium, and semibold weights, each in both normal and italic versions. The fonts are loaded from the Guardian’s servers in three different file formats: WOFF2, WOFF, and TrueType.@font-face {
font-family: Guardian Headline Full;
src: url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Bold.woff2) format(“woff2”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Bold.woff) format(“woff”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Bold.ttf) format(“truetype”);
font-weight: 700;
font-style: normal;
}
@font-face {
font-family: Guardian Headline Full;
src: url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BoldItalic.woff2) format(“woff2”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BoldItalic.woff) format(“woff”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BoldItalic.ttf) format(“truetype”);
font-weight: 700;
font-style: italic;
}
@font-face {
font-family: Guardian Headline Full;
src: url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Black.woff2) format(“woff2”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Black.woff) format(“woff”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-Black.ttf) format(“truetype”);
font-weight: 900;
font-style: normal;
}
@font-face {
font-family: Guardian Headline Full;
src: url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BlackItalic.woff2) format(“woff2”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BlackItalic.woff) format(“woff”),
url(https://assets.guim.co.uk/static/frontend/fonts/guardian-headline/noalts-not-hinted/GHGuardianHeadline-BlackItalic.ttf) format(“truetype”);
font-weight: 900;
font-style: italic;
}
@font-face {
font-family: Guardian Titlepiece;
src: url(https://interactive.guim.co.uk/fonts/garnett/GTGuardianTitlepiece-Bold.woff2) format(“woff2”),
url(https://interactive.guim.co.uk/fonts/garnett/GTGuardianTitlepiece-Bold.woff) format(“woff”),
url(https://interactive.guim.co.uk/fonts/garnett/GTGuardianTitlepiece-Bold.ttf) format(“truetype”);
font-weight: 700;
font-style: normal;
}
@media (min-width: 71.25em) {
.content__main-column–interactive {
margin-left: 160px;
}
}
@media (min-width: 81.25em) {
.content__main-column–interactive {
margin-left: 240px;
}
}
.content__main-column–interactive .element-atom {
max-width: 620px;
}
@media (max-width: 46.24em) {
.content__main-column–interactive .element-atom {
max-width: 100%;
}
}
.content__main-column–interactive .element-showcase {
margin-left: 0;
}
@media (min-width: 46.25em) {
.content__main-column–interactive .element-showcase {
max-width: 620px;
}
}
@media (min-width: 71.25em) {
.content__main-column–interactive .element-showcase {
max-width: 860px;
}
}
.content__main-column–interactive .element-immersive {
max-width: 1100px;
}
@media (max-width: 46.24em) {
.content__main-column–interactive .element-immersive {
width: calc(100vw – var(–scrollbar-width));
position: relative;
left: 50%;
right: 50%;
margin-left: calc(-50vw + var(–half-scrollbar-width)) !important;
margin-right: calc(-50vw + var(–half-scrollbar-width)) !important;
}
}
@media (min-width: 46.25em) {
.content__main-column–interactive .element-immersive {
transform: translate(-20px);
width: calc(100% + 60px);
}
}
@media (max-width: 71.24em) {
.content__main-column–interactive .element-immersive {
margin-left: 0;
margin-right: 0;
}
}
@media (min-width: 71.25em) {
.content__main-column–interactive .element-immersive {
transform: translate(0);
width: auto;
}
}
@media (min-width: 81.25em) {
.content__main-column–interactive .element-immersive {
max-width: 1260px;
}
}
.content__main-column–interactive p,
.content__main-column–interactive ul {
max-width: 620px;
}
.content__main-column–interactive:before {
position: absolute;
top: 0;
height: calc(100% + 15px);
min-height: 100px;
content: “”;
}
@media (min-width: 71.25em) {
.content__main-column–interactive:before {
border-left: 1px solid #dcdcdc;
z-index: -1;
left: -10px;
}
}The interactive content column has a left border and specific spacing for elements. Paragraphs following certain elements get extra top padding, and the first letter of these paragraphs is styled as a large, colored drop cap. Inline elements and figures are limited to a maximum width. Various color variables are defined for consistent theming.Pullquotes within specific containers should have a maximum width of 620 pixels.
Captions for showcase elements in main articles, features, standard articles, and comments should be positioned statically, span the full width, and also have a maximum width of 620 pixels.
Immersive elements should span the full viewport width, minus the scrollbar. On larger screens (over 71.24em), their maximum width is 978px, with caption padding adjusted at different breakpoints. On medium screens (between 46.25em and 61.24em), the maximum width is 738px. On smaller screens (under 46.24em), these elements should have no left margin and align to the left edge, with adjusted margins and caption padding for very small screens.
For the furniture wrapper on large screens (over 61.25em), a grid layout is used with defined columns and rows. Headlines get a top border, meta information is positioned relatively, and standfirst text has specific styling for links and spacing. Figures within this wrapper have adjusted margins and a maximum width of 630px for inline elements. On extra-large screens (over 71.25em), the grid template is adjusted to include more columns.The CSS defines a grid layout for a page header with areas for title, headline, standfirst, meta information, and a portrait. It sets specific dimensions and positions for these elements at different screen sizes. Styling includes borders, fonts, and colors, with adjustments for social media icons, captions, and responsive behavior on mobile and desktop views.The CSS code defines styles for a webpage layout, focusing on a component called “furniture-wrapper.” This component has a dark background and adjusts its margins and padding based on screen size. For larger screens, it adds decorative sidebars.
Inside this wrapper, elements like article headers, headlines, and meta information (such as social sharing buttons) are styled. Headlines use a light gray color and bold font. Social buttons have colored borders that match a theme color, and they change to a solid color on hover.
The code also manages the visibility of captions and a caption toggle button for media elements. Overall, the styles ensure the layout is responsive and visually consistent across different devices.This CSS code defines styles for a webpage component, likely a sidebar or content wrapper. It sets text colors, link appearances, and layout adjustments for different screen sizes. Links are styled with specific colors and underlines, and the layout includes background elements that adjust based on viewport width. The code also handles borders and spacing for various elements within the component.The CSS code defines styles for article elements and loads custom fonts. It sets the border color for comments to match the header border color. For article headings, it uses a light font weight by default, but switches to bold when the heading contains a strong tag.
Additionally, it imports the Guardian Headline Full font family in various weights and styles (light, regular, medium, semibold, each with normal and italic versions) from the Guardian’s servers, specifying multiple file formats for browser compatibility.The font “Guardian Headline Full” is defined with several styles: semibold italic (weight 600), bold (weight 700), bold italic (weight 700), black (weight 900), and black italic (weight 900). Each style includes font files in WOFF2, WOFF, and TrueType formats, hosted on the Guardian’s servers.
Additionally, the “Guardian Titlepiece” font is defined in its bold style (weight 700), also with WOFF2, WOFF, and TrueType files.
For iOS and Android devices, a dark background color and specific feature colors are set. In dark mode, the feature color adjusts accordingly.
On iOS and Android, specific CSS rules are applied to style the first letter of the first paragraph in certain article containers, following specific elements like `.element-atom` or sign-in gates.For Android devices, the first letter of the first paragraph in standard and comment articles is styled with a secondary pillar color. On both iOS and Android, article headers are hidden, and the furniture wrapper has specific padding. Labels within the furniture wrapper use a bold, capitalized font with a new pillar color. Headlines are set to 32px, bold, with bottom padding and a dark color. Image figures are positioned relatively, with full-width margins and automatic height.For Android devices, images within article containers should have a transparent background, span the full viewport width (accounting for scrollbars), and adjust their height automatically.
On both iOS and Android, the introductory text in articles should have specific spacing: 4 pixels of padding at the top and 24 pixels at the bottom, with a 10-pixel negative margin on the right.
The paragraphs within this introductory text should use the Guardian’s headline font family.
Links inside this text should be styled with a specific color, an underline positioned 6 pixels below the text, and a light gray underline color, with no background image or border. When hovered over, the underline should change to match the link’s text color.
Additionally, the metadata section in articles on both iOS and Android should receive specific styling.This CSS code sets styles for article containers on iOS and Android devices. It adjusts margins, colors, padding, and button displays for various article types and their components.On iOS and Android devices, for feature, standard, and comment article containers, images that are not thumbnails or immersive are styled to have no margin, a width calculated from the viewport minus 24 pixels and the scrollbar width, and an automatic height. Their captions have no padding.
Immersive images in these containers are set to a width that is the viewport width minus the scrollbar width.
Within the article body’s prose, quoted blockquotes display a colored marker using a custom CSS variable. Links are styled with a specific color, an underline offset by 6 pixels, and a decoration color from a variable. On hover, the underline color changes to another variable.
In dark mode, the background color of the furniture wrapper in these containers is set to a dark gray (#1a1a1a).For iOS and Android devices, the content labels in feature, standard, and comment articles use the new pillar color. Headlines in these articles have no background and use the header border color. The standfirst text and links, along with author bylines, also adopt the new pillar color. Icons in the meta section are styled with the same color. Captions for showcase images use the dateline color, and quoted text within the article body follows the same styling.For iOS and Android devices, the text color of quoted blocks within article bodies is set to a specific pillar color. Additionally, the background color for various article body containers is forced to a dark background. On iOS, the first letter following certain elements in these containers is styled with a drop cap effect.This appears to be a CSS selector targeting the first letter of paragraphs in specific article containers on iOS and Android devices. The selector applies to various article types (standard, feature, comment) and accounts for different page structures and sign-in gate elements.This CSS code sets styles for specific elements on Android and iOS devices. It defines colors, padding, margins, and other visual properties for various article containers, buttons, and text elements. The code also includes a dark mode media query that adjusts colors when the user prefers a dark color scheme. Some elements, like article headers, are hidden by setting their opacity to zero.The CSS code sets styles for article containers on iOS and Android devices. It removes margins from furniture wrappers and adjusts colors for labels, headlines, and links based on custom properties or specific values. Headlines are set to a light gray color, while links and labels use a variable color. Meta sections feature a striped background created with a repeating linear gradient, and bylines are also styled in light gray.For iOS and Android devices, links within the meta section of feature, standard, and comment articles should use the new pillar color (or a dark mode feature color as a fallback).
Similarly, SVG icons within the meta section’s miscellaneous area for these article types on both iOS and Android should have their stroke set to the same new pillar color.
Labels for alerts within the meta section across all these article containers on iOS and Android should be displayed in the color #dcdcdc, with this rule taking priority.
Additionally, any span elements containing a data-icon attribute within the meta section for these articles on both platforms should also adopt the new pillar color.For iOS and Android devices, the icon color within meta sections of feature, standard, and comment article containers is set to a specific color variable.
On larger screens, the meta section in these containers displays a top border and adjusts the margin for its miscellaneous content.
Additionally, paragraphs and lists within the article body of these containers have a maximum width. Blockquotes styled as “quoted” also receive specific styling.For Patrizia Schlosser, it began with an apologetic call from a colleague. “I’m sorry, but I found this. Are you aware of it?” He sent her a link that led to a site called Mr DeepFakes. There, she discovered fake images of herself—naked, squatting, chained, and performing sex acts with various animals. They were tagged “Patrizia Schlosser sluty FUNK whore” (sic).
“They were very graphic, very humiliating,” says Schlosser, a German journalist for Norddeutscher Rundfunk (NDR) and Funk. “They were also very poorly done, which made it easier to distance myself and recognize they were obviously fake. But it was deeply disturbing to imagine someone somewhere spending hours online searching for pictures of me, putting all this together.”The site was new to Schlosser, despite her previous high-profile investigations into the porn industry. “I’d never heard of Mr. DeepFakes—a porn site entirely dedicated to fake porn videos and photos. I was surprised by how big it was, with so many videos of every celebrity you know.” Schlosser’s first reaction upon seeing herself among them was to brush it aside. “I tried to push it to the back of my mind, which was really a strategy of not dealing with it,” she says. “But it’s strange how the brain works. You know it’s fake, but still you see it. It’s not you, but also it is you. There you are with a dog and a chain. You feel violated but confused. At some point, I decided: ‘No. I’m angry. I don’t want those images out there.'”
Schlosser’s subsequent documentary for NDR’s STRG_F program succeeded in getting the images removed. She also tracked down the young man who had created and posted them—even visiting his home and speaking to his mother. (The perpetrator himself wouldn’t come out of his bedroom.) However, Schlosser was unable to identify “Mr. DeepFakes,” or whoever was behind the site, despite enlisting the help of Bellingcat, the online investigative journalism collective. Bellingcat’s Ross Higgins was on the team. “My background is investigating money laundering,” he says. “I looked at the structure of the website, and it was using the same internet service providers (ISPs) as proper serious organized criminals.” The ISPs suggested links to the Russian mercenary group Wagner and individuals named in the Panama Papers. The ads it carried included ones for apps owned by Chinese technology companies, which allowed China’s government access to all customer data. “I made the presumption that this was all much too sophisticated to be a site of hobbyists,” says Higgins.
It turned out that’s exactly what it was.
The story of Mr. DeepFakes, the world’s largest, most notorious nonconsensual deepfake porn site, is really the story of AI porn itself—the very term “deepfake” is believed to have come from its originator. A “ground zero” for AI-generated pornography, its pages—which have been viewed more than 2 billion times—have depicted countless female celebrities, politicians, European princesses, and wives and daughters of U.S. presidents being kidnapped, tortured, shaved, bound, mutilated, raped, and strangled. Yet all this content (which would take more than 200 days to watch) was just the site’s “shop window.” Its true heart, its “engine room,” was its forum. Here, anyone wanting deepfakes created of someone they knew—a girlfriend, sister, classmate, or colleague—could find someone willing to make them to order for the right price. It was also a “training ground,” a technical hub where “hobbyists” taught one another, shared tips, posted academic papers, and problem-solved. (One recurring problem was how to deepfake without a good “dataset.” This means when you’re trying to deepfake someone you don’t have many pictures of—so not a celebrity, but maybe someone you know whose social media you’ve screengrabbed.)
The filmmaker and activist Sophie Compton spent many hours monitoring Mr. DeepFakes while researching the award-winning 2023 documentary Another Body (available on iPlayer). “Looking back, I think that site played such an instrumental role in the proliferation of deepfakes overall,” she says. “I really think that there’s a world in which the site didn’t get made, wasn’t allowed to be made, or was shut down quickly, and deepfake porn is just a fraction of the issue that we have today. Without that site, I don’t think it would have exploded in the way it did.”
In fact, that scenario…The origins of Mr Deepfakes date back to 2017-18, when AI-generated porn was just starting to gain traction on social media platforms like Reddit. An anonymous Redditor and AI porn “pioneer,” who used the name “deepfakes” and is credited with coining the term, gave an early interview to Vice about its potential. However, in early 2018, Reddit banned deepfake porn from its site. “We have screenshots from their message boards at that time, and the small deepfake community was panicking and looking for alternatives,” says Compton. This led to the creation of Mr DeepFakes, initially under the domain dpfks.com. The administrator, using the same username “dpfks,” advertised for volunteer moderators, posted rules and guidelines, shared deepfake videos, and provided a detailed guide on using software to create deepfake porn.
“What’s so disheartening about reading those messages and seeing how it all began is realizing how easily governments could have stopped this early on,” says Compton. “The people involved didn’t think they’d be allowed to continue freely. They were saying things like, ‘They’re coming for us!’ and ‘They’ll never let us do this!’ But as they faced no consequences, they grew bolder. The pandemic added to the explosion, as content moderation dropped. The content was violent and aimed at completely degrading individuals. Popular targets were often very young celebrities, like Emma Watson, Billie Eilish, and Millie Bobby Brown.” (Greta Thunberg is another example.)
Who was behind it? Mr DeepFakes occasionally gave anonymous interviews. In a 2022 BBC documentary, Deepfake Porn: Could You Be Next?, the site’s “owner” and “web developer,” using the pseudonym “deepfakes,” argued that consent from the women wasn’t necessary because “it’s a fantasy, it’s not real.”
Was money the motivation? Mr DeepFakes ran ads and offered a premium membership paid in cryptocurrency. In 2020, a forum mentioned the site was making between $4,000 and $7,000 a month. “There was a commercial aspect,” says Higgins. “It was a side hustle, but it was more than that. It brought notoriety.”
At one point, the site “posted 6,000 pictures of AOC’s [Alexandria Ocasio-Cortez’s] face so people could make deepfake pornography of her,” says Higgins. “It’s insane. There were files of YouTubers and politicians. What this says is that if you’re a woman in this world, you can only achieve so much. If you dare to step into the public eye, you can expect your image to be used in the most degrading way possible for someone else’s profit.”
“The most disturbing thing for me was the language used about women on that site,” he continues. “We had to edit it for our online report to avoid triggering readers, but it was pure misogyny. Pure hatred.”
In April, investigators believed they had identified Mr DeepFakes and emailed their suspect.
On May 4, Mr DeepFakes shut down. A notice on its homepage blamed “data loss” due to the withdrawal of a “critical service provider.” “We will not be relaunching,” it stated. “Any website claiming otherwise is fake. This domain will eventually expire, and we are not responsible for its future use. This message will be removed in about a week.”
Mr DeepFakes is now finished—but according to Compton, this could have happened much sooner. “All the signs were there,” she says. The previous year, in April 2024, when the UK government announced plans to criminalize the creation and sharing of deepfake pornography.In response to concerns about deepfake sexual abuse material, Mr DeepFakes promptly blocked access for users in the UK. (These plans were later put on hold when the 2024 election was announced.) “This showed that ‘Mr DeepFakes’ was clearly not so committed that governments were powerless,” says Compton. “If running the site was going to become too much trouble and too risky, they simply weren’t going to bother.”
However, deepfake pornography has now become so widespread and mainstream that it no longer needs a central hub or “base camp.” “The skills those individuals prided themselves on learning and teaching are now so commonplace that anyone can access them through apps with just a click of a button,” Compton explains.
For those seeking more sophisticated creations, the self-styled experts who once operated in the site’s forums are now openly advertising their services. Patrizia Schlosser experienced this firsthand. “As part of my research, I went undercover and contacted some people on the forums, requesting a deepfake of an ex-girlfriend,” says Schlosser. “Even though it’s often claimed the site was only for celebrities, that wasn’t true. The response was simply, ‘Yeah, sure…’
“After Mr DeepFakes shut down, I received an automated email from one of them saying: ‘If you want anything made, let me know… Mr DeepFakes is down—but of course, we’re still working.'”
Support resources:
UK and Ireland: Samaritans can be contacted on freephone 116 123, or by email at jo@samaritans.org or jo@samaritans.ie.
US: You can call or text the 988 Suicide & Crisis Lifeline at 988, or chat online at 988lifeline.org.
Australia: The crisis support service Lifeline is available at 13 11 14.
Other international helplines can be found at befrienders.org.
For support related to rape and sexual abuse:
UK: Rape Crisis offers support on 0808 802 9999 in England and Wales, 0808 801 0302 in Scotland, or 0800 0246 991 in Northern Ireland.
US: Rainn offers support on 800-656-4673.
Australia: Support is available at 1800Respect (1800 737 732).
Other international helplines can be found at ibiblio.org/rcip/internl.html.
Frequently Asked Questions
Of course Here is a list of FAQs about the Mr DeepFakes story designed to answer questions from basic to more advanced
Beginner Definition Questions
1 What was Mr DeepFakes and what does deepfake mean
Answer Mr DeepFakes was the username of the anonymous Reddit user who in 2017 created and shared a software tool that made it easy for people to create deepfakes A deepfake is a hyperrealistic AIgenerated video or image where a persons face or body is swapped with someone elses
2 What was the main purpose of the original Mr DeepFakes site
Answer Overwhelmingly it was used to create nonconsensual pornographic videos by superimposing celebrities and other peoples faces onto adult film actors bodies The tagline It was about completely degrading someone reflects its malicious intent
3 Why is this story such a big deal
Answer It marked a turning point where powerful AI image manipulation once limited to researchers became widely accessible It highlighted a terrifying new form of digital sexual abuse and harassment that is difficult to detect and prosecute
Impact Ethical Questions
4 Who were the victims of these deepfakes
Answer Initially the primary targets were female celebrities However the technology quickly spread to target noncelebrities including expartners colleagues and classmatesa form of abuse often called revenge porn 20
5 What happened to the original Mr DeepFakes site
Answer Due to massive public backlash and policy changes on platforms like Reddit the original forum and toolsharing hubs were banned and shut down in early 2018 However the software and knowledge spread instantly to other parts of the internet
6 Isnt this just a harmless prank or parody
Answer No Creating and sharing sexually explicit deepfakes of someone without their consent is a serious violation It causes profound psychological harm reputational damage and is a form of imagebased sexual abuse even if the underlying video is fake
Technical Legal Questions
7 How does the technology behind deepfakes actually work
