The best deepfake examples

deepfake of Jim Carrey as Jack Torrence in the Shining

(Image credit: Ctrl Shift Face)

The best deepfake examples show how quickly AI-driven face-swap technology has developed in recent years – and why the tech is controversial. The ability to map anyone's face over an actor's, or even to generate convincing AI video from a short clip or a single image, has huge creative potential for VFX, film and other visual arts, but it has also been used to mislead and to spread fake news.

Many deepfakes have been used for humour, creating parodies and memes, while we've also seen big brands embrace the tech for high-profile ad campaigns. To learn more about generative AI, see our roundups of the best AI image generators and the biggest AI art controversies of 2024. In the meantime, here are 25 milestones that demonstrate the advances in deepfakes video over the past years.

You may like

Created by Youtuber EZRyderX47 the video is a great deepfake example that shows the creative possibilities of the technology. As Marty, Holland is a surprisingly good stand-in for Michael J. Fox and fans seem to agree, with the video racking up a sturdy 11 million views since its release. Now all we need is a deepfake of Christopher Lloyd as Iron Man and Michael J. Fox as Spiderman – somebody make it happen.

“Hey y’all, it’s Taylor Swift here," the fake Taylor says. "Due to a packaging error, we can’t sell 3,000 Le Creuset cookware sets. So I’m giving them away to my loyal fans for free,” she adds. While those of us familiar with AI technology will instantly recognise her strangely robotic cadence, some fans were tricked out of money by the scam. Deepfakes are creepy enough without the added threat, so stay vigilant out there on the World Wide Web.

Another Tom Holland example - and this one's even stranger. This example is a clip from ITVX's Deep Fake Neighbour Wars – a show that's riddled with strange deepfaked celebs in bizarre settings. The video in question depicts rapper Nicki Minaj and actor Tom Holland as a distraught couple, recounting a home invasion by none other than Mark Zuckerberg. The odd clip was shared on X where it was seen by Minaj, who was rather disturbed, to say the least.

The video's still hugely impressive, and frightening, a year on, as we saw when it resurfaced on Twitter last month. "How can this tech NOT be deployed in the 2024 election?" one user commented. "Soon we’ll see that even this is essentially child’s play when it comes to the actual, ever-present (yet invisible) capabilities of identity manipulation and whole-cloth digital identity creation…the implications of which are far-reaching & bone chilling," someone added.

Bizarre film/actor crossovers are popular amongst deepfake creators; you can find plenty of them on YouTube (including the above Jerry Seinfield and Pulp Fiction mashup). This terrifyingly good edit sees Jim Carrey take on the role of Jack Torrence in a series of videos showing the most important moments from the 1980 film The Shining. It’s scarily convincing, and makes us want to see Jim Carrey staring in a horror film. His exaggerated expressions would be perfect.

This hilarious and perfectly executed creation sees DesiFakes puts Jerry Seinfeld into one of the most famous scenes from Pulp Fiction, and it works fantastically. The audio editing is as important as the visuals here, with perfect timing right down to the inappropriate canned laughter and the jingle and credit roll. The facial expressions could come straight out of a Seinfeld episode. This is the kind of things deepfakes were made for.

@deeptomcruise
♬ original sound - heyy

@unreal_keanu
♬ original sound - Unreal Keanu Reeves

As with Cruise above, this particular deepfake celeb is being used for laughs and isn't being passed off as real. But despite the humour, it shows the risks deepfakes pose for celebs. On the one hand, the tech could allow celebrities to sell their image to brands without having to go to the trouble of actually filming a spot or going to a photoshoot. But on the other side, there have already been reports of unscrupulous brands using deepfakes to push their products without the celebs' permission.

The channel warned viewers in advance that the newsreader would be faked, and Kim Joo-Ha still has her job. However, MBN said it planned to continue using the deepfake for some breaking news reports, and the company behind the deepfake, South Korea's DeepBrain AI (formerly known as Moneybrain), has said that it's looking for media buyers in China and the US, leading some to fear that newsreaders may become obsolete.

Remakes and reboots continue to be a massive part of the modern film landscape. Whenever new actors are cast in classic roles, comparisons are inevitably made between the different portrayals. Deepfake technology has allowed people to take those comparisons one step further, by putting one actor in the place of another for a sequence, highlighting similarities and differences that are incredibly interesting to observe.

The entertainment landscape is undergoing a seismic shift right now, as streaming services like Netflix battle with the big screen for attention. With this in mind, Collider put together this hilarious deepfake with the super recognisable faces of Tom Cruise, Robert Downey, Jr, George Lucas, Ewan McGregor and Jeff Goldblum discussing streaming and the future of cinema. This is a favourite of ours not only because it's incredibly convincing but also because it's a highly amusing video. As one commenter puts it, this is 'scary good'.

Many of the most convincing deepfake examples have been created with the help of impersonators that mimick the source’s voice and gestures, just like this video produced by BuzzFeed and comedian Jordan Peele using After Effects CC and FakeApp. Peele’s mouth was pasted over Obama’s, replacing the former president’s jawline with one that followed Peele’s mouth movements. FakeApp was then used to refine the footage through more than 50 hours of automatic processing.

[NEW RELEASE] ‘I wish I could...’ (2019) Mark Zuckerberg, founder of @facebook reveals the ‘truth’ about privacy on #facebook . This artwork is part of a series of AI generated video works created for ‘Spectre’ - an immersive exploration of the digital influence industry, technology and democracy. [link in bio]. Art by Bill Posters & @danyelhau for @sheffdocfest @sitegallery #artnotmisinformation #thefutureisntprivate #deepfakes #deepfake #spectreknows #surveillancecapitalism #privacy #democracy #dataism #contemporaryartwork #digitalart #generativeart #newmediaart #codeart #markzuckerberg #artivism #contemporaryart Bill Posters
A photo posted by on

Instagram didn’t take the Zuckerberg video down, but said it would, “treat this content the same way we treat all misinformation on Instagram. If third-party fact checkers mark it as false, we will filter it.” The posters had flagged it using the hashtag #deepfake. While the video is reasonably convincing on mute, the voice gives it away, showing that a good actor is still needed to make plausible deepfake examples (although AI voice synthesis has been advancing by leaps and bounds, so perhaps not for long).

Trump again! This one's an old one, but it's notable for having been the first example of a political party using a deepfake, Belgium’s Socialistische Partij Anders (sp.a) posted this video on Facebook back in May 2018 showing Trump taunting Belgium for remaining in the Paris climate agreement. With Trump’s hair looking even stranger than usual and the crude movement of the mouth, it’s very clearly fake, and the voiceover says as much, though the final line “We all know that climate change is fake, just like this video,” isn’t subtitled in Flemish, but it was still enough to provoke one user to comment “Trumpy needs to look at his own country with its crazy child killers,” and for sp.a to have to clarify it was fake.

Here's an example of deepfakes being used by a major brand, and for a marketing campaign with a positive mission. Dove used deepfake technology to put very unlikely advice into the mouths of the mothers of teenage girls, the aim being to raise awareness of the negative impact of a lot of the dangerous 'beauty' advice shared by influencers on social media apps. In the campaign video, the participants sit mouths wide, clearly horrified by the distorted advice being given by their mums. Entitled Toxic Influence, the ad was created by Ogilvy.

If a great impersonation is the basis of a convincing deepfake then this video of actor and comedian Bill Hader morphing into Hollywood legends Al Pacino and Arnold Schwarzenegger couldn't fail. Although the team at Ctrl Shift Face aren't trying to trick anyone with this video, they do demonstrate how crucial it is to have an actor that can capture the mannerisms of your source. The subtle morphing of Hader's face into that of Pacino and then Schwarzenegger is slightly terrifying and completely hilarious.

Agency GS&P pulled off the kind of headline-grabbing stunt that the publicity-loving Dalí would have appreciated when they resurrected the Catalan artist as a charismatic host at the Dalí Museum in Florida. Billed as “art meets artificial intelligence”, Dalí Lives was created by pulling more than 6,000 frames from old video interviews and processing them through 1,000 hours of machine learning before overlaying the source onto an actor’s face. The text was comprised of quotes from interviews and letters with new commentary designed to help visitors empathise with the artist and relate to his work.

The novelty of this deepfake example is its interactivity. A total 45 minutes of footage split over 125 videos allows for more than 190,000 possible combinations depending on visitor responses and even includes comments on the weather. It finishes with Dalí turning around and snapping a selfie with his audience. Dalí claimed it was unlikely he would ever die, and maybe he was right, because he was brought to life a second time recently by Samsung’s AI lab in Moscow, this time by training AI on landmark facial features from just a handful of images rather than the usual thousands.

Fortunately, most known deepfake examples online are clearly flagged as fake and are not intended to fool anyone. They're usually being played for laughs, for example, putting Nicolas Cage in every film ever produced, or at worst for sordid fantasy (deepfake celebrity porn through which the tech first took root has been banned). But there have been cases of people using deepfakes for nefarious ends and trying to pass them off as real.

Its creator, a fan of Yang Mi, issued an apology on microblogging site Weibo and said he’d made the video as a warning to raise awareness of the technology. We can actually see lots of possible uses of deepfakes for the film and television industry. It’s also possible to see how the industry could eventually embrace the technology and turn it to profit by allowing viewers to play director on home releases, manipulating dialogue, incorporating alternative scenes or even inserting themselves as characters. We won't be surprised if we see a lot of video games with celebrity appearances too.

What is a deepfake?

A deepfake is essentially an image, video or audio recording in which the image or voice of a human has been faked digitally using tools that make use of machine learning algorithms or artificial intelligence. The technology can be used to change the voice or appearance of an actor to make them resemble someone else, be it a real-life person or a fictional character.

Related technology has been used in the film industry to transpose real or fictional faces onto other actors, for example, bringing Peter Cushing back to life in 2016’s Rogue One: A Star Wars Story. Such techniques required complex and expensive pipelines with face-mounted cameras, but more accessible software has since made deepfakes relatively easy to create.

How do deepfakes work?

Deepfakes take their name from the fact that they use deep learning technology, a kind of machine learning that applies neural net simulation to massive data sets. Artificial intelligence (AI) effectively learns the characteristics of a particular face at different angles in order to be able to recreate it. Deepfake video technology makes use of this to transpose a face onto a target as if it were a mask. Using an actor as a target generates more realistic movement than using straight AI-generated video.

Big advances came through the use of generative adversarial networks (GANS). These pit two AI algorithms against each other, one creating the fake and the other grading its efforts, teaching the synthesis engine to make more convincing fakes. Initially, deepfakes required a large amount of training data, which made celebrities the most common targets since there are often many easily available images of them to use. However, advances in AI video generation mean that it's now possible to create deepfakes from a short clip or even a single image.

Why are deepfakes controversial?

The technology behind deepfakes offers many interesting possibilities for various creative sectors, from dubbing and repairing video to solving the uncanny valley effect of CG characters in films and video games, avoiding actors having to repeat a fluffed line and the creation of apps that allow us to try new clothes and hairstyles. The technology is also being used to produce corporate training videos and train doctors.

However, there remains a widespread fear about the technology being used for more sinister ends. The ability to create a realistic video showing the likeness of famous people means that deepfakes can be used to create fake news stories and spread disinformation online. There are also fears about them being used for impersonation in scams and to create pornographic material using people's likeness without their consent. The most recent development to cause concern is the ability to create live deepfake live steams on the fly, allowing anyone to turn up to a Zoom meeting looking like someone else.

If you'd like to delve deeper into these concerns, check out our piece on the ethics of digital humans.

What is the law on deepfakes?

Like other forms of artificial intelligence, the technology behind deepfakes has advanced so fast in recent years that legislators have been slow to catch up, and many countries have no specific laws on deepfakes. However, there have been developments that aim to bring some protections.

In the US, some individual states have introduced legislation. For example, Texas SB 751 makes it a criminal offence to create a deceptive video with intent to damage the reputation of a candidate or influence the outcome of an election. And several states have legislation banning the creation of explicit deepfakes involving minors.

At the federal level, the DEFIANCE Act of 2024, if approved, would granted rights to relief for people affected by non-consensual activities involving intimate digital fakes, while the Protecting Consumers from Deceptive AI Act would require the National Institute of Standards and Technology to create task forces to help develop technical standards and guidelines for the identification of content created by generative AI. In the UK, making a sexually explicit deepfake is now a criminal offence under the Online Safety Act.

How can you recognise a deepfake video?

There are fears that convincing deepfakes could be used for criminal purposes or to trick people into believing fake news stories, so a lot of people are inevitably wondering how deepfakes can be spotted. There are organisations that work to validate the authenticity of videos and images shared online using various techniques, and there are apps, such as Siwei Lyu's open-source DeepFake-o-meter, which can be used to help detect fake videos.

However, there are also a few aspects that can help you detect deepfakes just by looking at them. Deepfake videos often have unnatural eye-blinking with pixel artifacts, and the edges of the subject's face often look pixelated. Look out for changing shadows too, since lighting can often be inconsistent in AI-created video.

Also pay close attention to the subject's lips to see if they genuinely match the audio, particularly on bilabial plosives like a 'b' and 'p' sound, which require the lips to close completely. Hair and facial hair can also be a giveaway in deepfakes.

If you find yourself on a video call with someone who you suspect might not be real, it could be a good idea is to ask the person to turn to the side. The software used to estimate facial poses for deepfake videos often assigns more landmarks to the front of a face and doesn’t do so well at acute angles seen from the side. In many cases, there are also fewer images of people in profile to use as training data.

Finally, always consider how plausible a situation is. With the amount of AI content now circulating online, it makes sense to maintain a healthy skepticism if an image or video seems unlikely.

Thank you for reading 5 articles this month* Join now for unlimited access

Enjoy your first month for just £1 / $1 / €1

*Read 5 free articles per month without a subscription

Join now for unlimited access

Try first month for just £1 / $1 / €1

TOPICS

Joe is a regular freelance journalist and editor at Creative Bloq. He writes news, features and buying guides and keeps track of the best equipment and software for creatives, from video editing programs to monitors and accessories. A veteran news writer and photographer, he now works as a project manager at the London and Buenos Aires-based design, production and branding agency Hermana Creatives. There he manages a team of designers, photographers and video editors who specialise in producing visual content and design assets for the hospitality sector. He also dances Argentine tango.

With contributions from

Abi Le GuilcherEcommerce Writer

Images showing AI face-swap technology putting Keanu Reeves in a scene from the TV series Severance

Viral video of Keanu Reeves in Severance shows the power of Tencent's new AI video generator

The 10 most read AI art stories of 2024

Cartoon of young person looking at the sun rise over a futuristic cityscape

Should we be scared or excited? What to expect from AI in 2025

Fake AI pictures of Brad Pitt used to scam a woman

This deepfake Brad Pitt scam is a cautionary tale on the dangers of AI catfishing

Digital art trends; a mozaic of images on Cara

10 must-know digital art trends for 2025

The biggest 3D art trends we'll see in 2025

Latest in AI Art

Do you think that Fantastic Four poster was made with AI? Join the conversation

An image of a giant woman with long orange hair in a spacesuit, on sale in an AI art auction at Christie's

Christie's first AI art auction provokes fierce debate – and it's hard to pick a side

An image of a person struggling to find good stock images on a stock image site that has been made unusable by AI. Logos of stock images sites Adobe Stock, Pexels, Vecteezy, Freepik and Creative Market

Designers say AI is making stock image sites unusable

An image of a chef looking in a fridge that may be AI generated

Netflix gets roasted over dodgy AI art (again)

The biggest AI art controversies of 2024

AI-generated images from AI Art Magazine

New AI art magazine shows its time to take the medium seriously (even if it's hilarious)

Latest in Features

Cartoons of the 1960s; A vibrant illustration of the Yellow Submarine, a psychedelically patterned background surrounds the iconic vessel.

Cartoons of the 1960s – the best animated movies feature the work of Chuck Jones, who ensured characters were "not realistic, but believable"

Wix vs Wix Studio: which one is better?

"AI is a reflection of human creativity": a day in the life of Kae Neskovic

ImagineFX Art Challenge is back!

Wheel Wold developer interview; a person rides a bike down a trial in a colourful breezy cartoon world

How Wheel World's massive rolling landscapes were made possible with Houdini

6 reasons not to overlook the Mac mini

25 of the best deepfake examples that terrified and amused the internet