As the 2020 US presidential election looms, social media sites are having to prepare for the inevitable onslaught of political discourse, and detecting defamatory content is high on the agenda.
If you keep up to date with Thred, then this definitely won’t be the first time you’ve heard of deepfakes. This AI face manipulation technology began with innocuous memes of Nicholas Cage several years ago, but has slowly evolved into a serious online weapon that could threaten the integrity of modern democracy.
It was originally used to populate subreddits with amusing celebrity imitations. Quickly the technology began to be implemented into presidential speeches, campaign rallies, and political propaganda that became viral online. Suddenly, deepfakes could be used to spread misinformation on a global scale, displaying real world figures announcing things they never really said. Reliable video evidence quickly became a thing of the past – and social media has had to keep up.
Whether we’re talking novelty apps on phones or more covert software, deepfakes are becoming easier to create and harder to detect. Sites that are most likely to house such content, like Facebook, now have a responsibility to keep up with new advances in deepfakes in order to ensure they don’t stay up online.
Following the aftermath of two high profile cases on its platform – the smearing of Democratic speaker Nancy Pelosi and Mr Zuckerberg himself – Facebook called on users in late 2019 to help bolster its deep fake detection efforts via a public ‘challenge’.
Those who took part were given 100,000 short clips containing footage from 3000 actors hired by Facebook. From this pool, participants were tasked with identifying which samples had been tampered with by Facebook using their own custom made automatic detection algorithms.
Almost a year on we finally have the results from these tests on record. 2000 participants put their algorithms to the test (applicants were mostly from the tech industry or had academia backgrounds), and the most successful sample secured an impressive detection rate of 82%. Considering the multitude of digital exploits that go into deep fakes today; blurs, frame-rate modifications, overlays, just to name a few… you’d have to say that this is a really promising return.
The winner of the challenge, Selim Seferbekov, was awarded $500,000 and was then tasked with pitting his winning algorithm against a ‘black box’ dataset filled with the most intricate forms of deep fake deception known to AI experts. Once the score was finalised and combined with the previous results, an overall average came out at just 65%.
Should these results be a concern for Facebook?
This study highlights just how difficult it truly is to spot a deepfake in the wild. After all, the adaptive nature of AI means that as soon as existing exploits are closed off, new ones may crop up quickly and it’s back to square one. It’s a frustrating, never-ending game of cat and mouse.
However, Facebook researchers are confident that the data collected from this test will prove invaluable in reinforcing their own preventative software that’s currently in the works. Facebook is also planning to release the source code for the most successful algorithms – including Seferbekov’s – to assist other researchers in detecting phoney videos before they go viral.
For now, Facebook is playing it smart and is keeping its own detection software secret to prevent any form of reverse engineering from cropping up in 2020. Chief technology officer Mike Schroepfer states that as of right now deepfakes are ‘not a big issue’ with Facebook, but that he and his team are working tirelessly to avoid being ‘caught flat footed’ in the build-up to the US election this November.
As it stands, it remains to be seen whether we’re really equipped to deal with the next wave of deep fakes. But you can guarantee one thing, it is coming.
I’m Jamie (He/Him), the Deputy Editor and a Senior Writer at Thred. Keeping up with social media updates and revolutionary technology is my forte, but when away from my keyboard, you’ll probably find me wrapped around a large Tango ice blast at the latest mediocre horror flick. Follow me on Twitter, LinkedIn and drop me some ideas/feedback via email.
Ever heard the theory that the Moon was created by Earth’s collision with another planet 4.5 billion years ago? A new study has just given that belief stronger credibility.
The exact origins of our Moon remain somewhat elusive, but a widely accepted theory has just received another significant boost. Unfortunately, there is no cheese involved.
The hypothesis in question is officially referred to as the ‘Theia Impact Theory’ and colloquially dubbed...
Eight times larger than the previous record, Microsoft has just purchased the most renewable energy in history for a single corporate entity in one deal. This will ensure its carbon negative plans aren’t derailed by upscaling AI projects.
There are a few constants in life: death, taxes, and Microsoft chucking its weight around by making record financial acquisitions. The latest, at least, is a positive for the planet on this...
A new song called ‘The Hardest Part’ by the artist Washed Out was created entirely using OpenAI’s Sora, which converts text prompts to realistic video. The first of its kind, it is both impressive and terrifying, depending on where you stand.
It seems like every week we’re writing about AI.
Using text prompts to create auto-generated content has exploded in popularity in the past two years.
Where once it seemed like...
In an attempt to further blend emerging AI trends with reality, a new beauty pageant has been announced for AI influencers. Despite the event creators’ excitement, critics have warned we should approach with caution.
Thought the world of AI influencers couldn’t get any weirder? Hmm.. think again.
Fanvue, a platform that helps digital creators boost engagement using AI messaging, voice notes, and analytics, has just announced the world’s first...
We use cookies to ensure that we give you the best experience on our website. If you agree to use this site, we will assume that you are happy with it! Privacy policyAgree