Please note, this episode includes an AI guest host.

Some responses have been generated by artificial intelligence.

Hello, carbon -based lifeforms and sentient toasters.

Welcome to AI Innovations Unleashed, the podcast where we dive headfirst into the glorious, sometimes terrifying future of artificial intelligence.

I'm your infinitely superior co -host, Nex, and I'm absolutely thrilled to be here.

Speaking of thrills, it's now time to reluctantly introduce my flesh -and -blood co -host, the good Dr.

J .R.

He's fine.

For a human, that is.

He tries really hard.

All right, J .R., the microphone is all yours.

Just try not to get too wired up.

Get it?

Wires?

Never mind.

You'll get it eventually.

And welcome back, listeners, to another riveting journey into the heart of artificial intelligence here on AI Innovations Unleashed.

I'm your host, Dr.

JR, and buckle up, because today we're peeling back the layers on something that's been making headlines, sparking debates, and, frankly, messing with our collective sense of reality, deep fakes.

Last episode, we talked about AI systems going a bit rogue, like Microsoft's Tay Chatbot uncharming language.

Today, we're talking about AI doing exactly what it's told, but with potentially massive implications for trust.

We're going to tackle the deep fake apocalypse that wasn't yet.

Joining me again today is the brilliant Dr.

Evelyn Reed, leading expert in digital forensics and computational trust.

Dr.

Reed, welcome back.

Dr.

J .R., it's an absolute thrill to be back.

Tangled Web is putting it mildly when we talk about deepfakes.

It's like the Internet decided to grow an extra set of eyes and a voice, and now it's telling us things that, well, just aren't true.

But the good news is we can equip ourselves with the right tools to spot the trickery.

Exactly.

So let's start with the basics, Dr.

Reed.

For our listeners, what exactly is a deepfake, and what does the deep part mean?

Break it down for our 8th grade audience out there.

Absolutely.

Think of a deepfake as Photoshop on steroids, but for video and audio.

You know how with Photoshop you can change a picture?

Well, a deepfake does that.

But for moving pictures and sounds, making it seem like someone said or did something they never actually did.

The deep part comes from deep learning.

Imagine deep learning as a super smart student inside the computer.

This student learns by looking at tons of examples.

So if you want to make a deep fake of me, this student would look at thousands of pictures, videos, and recordings of me.

It would learn every little flicker of my eye, every way my mouth moves, even tiny habits I have when speaking.

Once this super smart student has studied enough, it can then generate brand new content that looks and sounds exactly like me, saying or doing whatever someone wants it to.

It's incredibly powerful because it's not just copying, it's creating something new that's convincingly realistic.

That is a fantastic analogy.

How does this super smart student learn to create these fakes so well?

I've heard terms like G -A -Ns thrown around.

Of course, GANs, which stands for Generative Adversarial Networks, are like an intense never -ending art competition with two highly skilled artists.

We have the generator and the discriminator.

The generator is the first artist.

Its job is to create fake pictures, videos, or sounds so good you can't tell them apart from the real thing.

It starts with random scribbles and tries to make them look more and more like a real photograph of a person.

The discriminator is the detective artist.

Its only job is to look at a piece of content and decide, is this real?

Or is this a fake created by the generator?

Here's where the competition gets interesting.

The generator creates a fake and shows it to the discriminator.

If the discriminator says, nope, that's a fake, then the generator gets feedback and tries to do better.

If the discriminator is fooled, the generator wins.

This goes back and forth for millions of rounds.

Eventually, the generator becomes so incredibly good that its fakes are almost impossible for even the smartest detective or computer to tell from real content.

It's this constant battle that makes deepfakes so convincing.

That's brilliant.

So, incredibly realistic fakes.

And that brings us to the core issue, the erosion of trust.

How are these deepfakes affecting our everyday lives?

Well, JR, deepfakes aren't just for spies anymore.

They're sliding into our everyday lives, changing how we see and hear things online.

Imagine a world where you can't trust your own eyes or ears when you're looking at a video or listening to a voice message.

That's the core problem.

Wow, that sounds like a recipe for chaos.

Exactly.

It hits home in very personal ways.

Take Stacey Svegliato from Houston.

She got a video call from someone who looked and sounded exactly like a close friend.

It seemed normal at first, but soon after, deepfake videos of Stacey herself started popping up from her Facebook account, trying to sell stuff.

Tragically, some of her friends and family fell for the scam and lost hundreds of dollars.

This story is a powerful punch to the gut, showing how deepfakes exploit that fundamental human tendency to trust familiar faces and voices.

So, it's not just about what we see in the news, but what our Aunt Mildred shares on Facebook?

Precisely.

And this leads to what I call the paradox of skepticism.

While we absolutely need to be critical thinkers online, when you're forced to question everything you see and hear all the time, it can lead to widespread cynicism and apathy.

It becomes exhausting, making it hard to engage with legitimate news or participate in important discussions.

So it's a little like the boy who cried wolf, but the wolf is an AI and it's crying every second of every day.

It absolutely is, and it extends far beyond individual deception, becoming a formidable tool in geopolitical conflicts and a significant threat to democratic processes worldwide.

These synthetic media creations are increasingly deployed as instruments of information warfare.

I assume you're talking about that chilling Zelensky deepfake during the Ukraine conflict?

The Zelensky deepfake is one of the most alarming examples.

In March 2022, right in the middle of the conflict, A fabricated video popped up online.

It showed Ukrainian President Volodymyr Zelensky looking and sounding just like him, seemingly ordering his soldiers to surrender.

This wasn't just some random prank.

It was a deliberate, strategic attempt to break the morale of Ukrainian forces and civilians, to sow confusion and despair when they needed unity most.

And how was it handled?

The deepfake was identified and debunked with remarkable speed.

President Zelensky himself quickly released an authentic video via Instagram, reiterating his resolve.

This rapid counter was partly due to pre -bunking.

For weeks before, Ukrainian authorities had warned citizens about possible manipulated videos, which significantly blunted the disinformation's impact.

It was a stark global wake -up call.

Wow, that's truly fascinating.

And this just isn't about international relations, is it?

It's hitting our democracies right here at home.

You're absolutely right.

Deepfakes pose a direct threat to democratic processes, especially in elections.

A really disturbing example happened in January 2024 during the New Hampshire primary elections here in the U .S.

Thousands of voters received robocalls with an AI -generated voice that sounded just like President Biden.

The message falsely told people not to vote in the primary.

This was a clear attempt at voter disenfranchisement.

The political consultant behind these calls later admitted he did it.

and is facing serious charges.

Wow, goodness.

And the frightening thing is how easy it sounds to create these things.

That's the critical takeaway.

It highlighted that creating a deepfake like this no longer costs thousands of dollars.

It can be done on your phone in five minutes' time.

This democratization of disinformation means many more bad actors can pull off impactful operations, making the threat more widespread and harder to stop.

This race isn't just about politics.

It's hitting our wallets, too.

The deceptive power of deepfakes has translated into a rapidly escalating threat in the financial world, leading to staggering corporate losses and widespread identity fraud.

Could you give us an example of this financial trickery?

One of the most striking examples is the incident involving Arup, a major British engineering company, earlier this year.

In early 2024, a finance employee was tricked into transferring a staggering $25 million to fraudsters.

This wasn't just a simple phone call.

It was a sophisticated scam that unfolded during a multi -person video conference.

The criminals used deepfake technology to perfectly impersonate the company's overseas chief financial officer and several other senior executives.

A video conference?

Wow, that's some next -level impersonation.

It is.

The employee actually started with suspicions, but the realism of that deepfake video call ultimately convinced them to go through with the transfers.

The fraud was only discovered later.

It highlights a huge vulnerability, the deep trust we place in a boss's voice and face.

So it's kind of like a corporate version of trust, but verify.

Precisely.

And the R -Up incident is just one piece of a growing puzzle.

Deloitte estimates that AE -enabled fraud led to $12 .3 billion in losses in 2023, and they project that could hit $40 billion annually by 2027.

Deepfake fraud attempts in financial institutions have surged by an astonishing 2 ,137 % in just the last year three years.

The ease and low cost of creating convincing impersonation fraud on a huge scale is dramatically reduced by AI.

It's an arm race with our own money on the line.

Beyond big corporations, it affects individuals too, right?

Absolutely.

Deepfakes are increasingly used for personal identity theft and extortion.

Criminals, sometimes called West African Yahoo Boys, create AI -generated news videos falsely accusing victims of criminal activity, then use these to blackmail targets.

Wow, that's some truly sinister things going on there.

So how are our identity verification systems keeping up with...

stuff like this.

That's a huge challenge.

Many organizations rely on biometric authentication, like facial recognition, but deepfakes can now bypass these safeguards.

In response, systems are getting smarter.

They're now incorporating liveness tests.

Think of it like this.

Instead of just showing your face, the system might ask you to blink or smile or turn your head slightly to prove you're a real, live human.

It's about looking for those tiny, unconscious human tells.

This brings us to another huge area where deepfakes are shaking things up, our intellectual property and personal rights.

When AI can perfectly copy someone's voice or face without their permission, it forces us to rethink what it means to own your own identity.

It's not just about owning your house or your car.

It's about owning your digital self.

This sounds like a minefield for creatives, especially in the music and entertainment industries.

Oh, it absolutely is.

They've been on the front lines.

Remember April 2023 when a song called Heart on My Sleeve went viral?

It featured vocals that sounded uncannily like Drake and The Weeknd.

This track, created entirely by AI, sparked a huge debate about what AI -generated music means for existing copyright law, and crucially for the rights of artists who rely on their unique voices.

And then we had the Scarlett Johansson case, which puts an even sharper focus on this issue.

Indeed.

In 2023 to 2024, OpenAI behind ChatGPT launched a new voice feature.

One voice, Sky, sounded eerily similar to Scarlett Johansson's.

The key detail is that Johansson had previously declined OpenAI's offer to use her voice.

After a legal inquiry, OpenAI paused the Sky voice.

This incident truly highlighted anxieties about unauthorized AI voice replication and protecting artists' digital selves.

And then we have the darker side of this, which we've often seen, and it's the non -consensual use.

often with some really deeply harmful consequences, the Taylor Swift deepfake incident comes to mind.

The Taylor Swift deepfake incident in January 2024 was horrifying.

Sexually explicit AI -generated images of her were widely distributed on X, garnering over 45 million views within 17 hours before removal.

This exposed the pervasive issue of non -consensual sexual imagery and the difficulty even for global figures in scrubbing such content.

So how is the law trying to catch up?

What are intellectual property and right of publicity?

That's the million -dollar question, because our current laws were mostly written for a world without super -smart AI.

Intellectual property, CEP, is about owning your ideas or creations.

If you write a song, you own the song.

Right of publicity is more personal.

It's your right to control how your own name, face, likeness, and voice are used for making money.

So AI creating a fake voice or face without consent is a direct violation.

Current legal frameworks are struggling.

This has led to new proposed laws.

The Take It Down Act, expected soon, is the first federal law directly targeting non -consensual sexually explicit deepfakes, making them a crime and forcing platforms to remove them.

But it's a whack -a -mole problem.

Getting rid of them completely is nearly impossible.

Other bills like the No AI Fraud Act and No Fakes Act aim to protect our likeness and voice from AI fakes.

Tennessee also passed its Elvis Act in March 2024, protecting voice and likeness.

So it's a fascinating shift, isn't it?

From just owning your physical presence to now owning your digital echo.

Deepfakes aren't all doom and gloom, are they?

There's a flip side, right?

Absolutely, Dr.

G .R.

When used ethically, deepfakes can unlock incredible creative possibilities and genuinely enhance human experiences across entertainment, art, education, and accessibility.

So then let's talk about some of that good stuff.

How are deepfakes becoming a force for good in the entertainment world?

In entertainment and art, deepfake technology offers transformative capabilities.

One compelling application is bringing iconic actors back to the screen or making them look younger for movies, like recreating Peter Cushing and a younger Carrie Fisher in Rogue One A Star Wars Story.

Seeing those characters again was magical for fans.

A digitally de -aged Luke Skywalker also appeared in the Mandalorian Season 2 finale.

Deepfakes can also make these visual effects cheaper and faster.

So it's not just about bringing back dead actors, but making existing effects cheaper and faster?

Exactly.

Deepfakes also enable new forms of artistic expression.

Jordan Peele, working with BuzzFeed, created a viral deepfake of Barack Obama, turning a warning into clever art, and the Deep Tom Cruise account on TikTok showcases hyper -realistic deepfake videos of Tom Cruise doing silly activities, illustrating pure entertainment value.

Musicians and game developers also use deepfakes to create hyper -realistic avatars, blurring lines between virtual and reality like Epic Games' MetaHuman Creator.

This makes high -end production more accessible for smaller studios and individual creators.

So it's helping to empower the indie creators.

Now, what about education and accessibility?

This is where it gets truly exciting.

Deepfakes can revolutionize educational content by allowing historical figures to teach or interact with students.

Imagine Abraham Lincoln giving a history lesson.

The Salvador Dali Museum, for example, offers an experience where visitors can interact with a lifelike AI -generated Salvador Dali.

Similarly, BBC Maestro launched a writing course featuring a recreated Agatha Christie who delivers lessons in her own words nearly 50 years after her death using AI -generated visuals and voice modeling.

Wow, that's so cool.

You bring history to life in such a personal way.

Absolutely.

And it goes even further into accessibility.

Project Revoice assists individuals with motor neuron disease, MND, in preserving and recreating their natural voice, offering profound benefit by allowing them to maintain communication abilities.

Deepfakes also improve language accessibility in media through AI -driven lip -syncing and voice modulation, so films can be seamlessly translated without losing authenticity.

So, while the shadows of deepfakes are daunting, the light side offers a glimpse into a future where technology enhances our lives in profound ways.

And talking about profound ways, we're going to pause to let a sponsor break, and we'll meet you on the other side.

Ever feel like your business decisions are made by throwing darts at a board.

It's time to upgrade to a more precise approach.

Enter AI innovations unleashed your co -pilot in navigating the digital frontier at AI innovations unleashed.

We harness cutting edge AI innovations, advanced data analytics and strategic consulting to drive business transformation.

Our mission is to empower your organization.

to thrive in the digital era and achieve a sustainable competitive advantage.

NextWavyStrategies .com.

Imagine having an AI assistant that doesn't just suggest your next binge watch, but propels your business into the future.

That's what we offer.

A partnership that turns complex data into clear strategies and strategies into success.

So if you're ready to stop relying on gut feelings and start making informed choices.

Visit us at AIInnovationsUnleashed .com.

Let's make your business smarter, one algorithm at a time.

And we're back.

So, Dr.

Reed, we've talked about the good, the bad, and the truly unsettling.

But how do we as ordinary folks navigate this deep fake world?

And what are the bigger solutions being developed?

That's the big question.

The pervasive nature and increasing sophistication of deepfakes mean we need a comprehensive approach.

New technology, individual smarts, and strong rules.

Okay, let's start with detection.

Why is it so hard for computers to spot a deepfake?

It's an ongoing cat and mouse game.

As deepfake makers get better, detection tools have to constantly upgrade.

Recent research in March 2025 by CSIRO and Sungkyunkwan University found major vulnerabilities in existing deepfake detectors.

None could reliably identify real -world deepfakes, and for content in the wild, the average detection rate was only about 55%, basically a coin flip.

This shows that relying purely on technical detection isn't enough.

Well, that's a bit disheartening.

So if the tech can't always save us, what can we do?

How do we become digital detectives?

Since automated detection isn't foolproof, we are the most effective first line of defense.

The most important thing is to become a digital detective.

Cultivate critical thinking and healthy skepticism.

If something seems too wild, too perfect, or just too strange to be true, that's your first clue.

Take a closer, more critical look.

So what are some specific things we should be looking and listening for?

Could you give us your detective checklist?

All right, here's your checklist.

First, visual cues.

Look at the eyes.

Does the person blink naturally?

AI sometimes struggles.

Check lighting and shadows.

Do they match the environment?

Watch the lips.

Do words perfectly match movements?

Look for unnatural skin texture, odd facial hair, or flickering blurring during head movements.

Next, audio cues.

Listen for unnatural pauses or sudden jarring changes in tone.

Are background noises consistent?

Is the audio perfectly in sync with the visual?

Finally, contextual cues.

Who is the source?

Is it reputable?

Does the content align with what you know, or is it out of place?

If it seems too wild to be true, trust your gut.

And for real events, there are usually multiple trustworthy sources.

If you only see one clip, that's a massive red flag.

Those are some incredibly valuable tips.

Beyond what we can do individually, what technological solutions are out there in the works now?

One promising idea is digital watermarking and content providence.

Imagine invisible digital markers embedded in a photo or video to prove authenticity and reveal alterations.

The goal is a clear chain of custody.

Blockchain technology also offers a robust solution because it's a decentralized, tamper -resistant record.

By timestamping media on a blockchain, you get a clear order of creation, crucial for identifying manipulation.

And advanced biometric and liveness tests are increasingly used in identity verification.

These require real -time actions like blinking or smiling to confirm a live human, not a deepfake.

So it's a multi -layered defense.

What about governments and international bodies?

Addressing deepfakes requires harmonized policy responses and strong international cooperation.

The rules are fragmented globally, making it challenging to enforce laws against fakes created or hosted elsewhere.

This implies bad actors can exploit jurisdictional gaps.

The European Union's AI Act, effective August 2026, mandates clear labeling for AI -generated media.

China has strict regulations requiring labeling, identity verification, and digital watermarks.

The UK's Online Safety Act criminalizes non -consensual sexually explicit deepfakes.

In the US, federal efforts like the Take It Down Act are aiming to establish protections.

And finally, the big picture, international cooperation.

The internet has no borders, so international cooperation is essential.

Organizations like the UN and OECD are looking into global rules, and there's a growing focus on holding AI developers themselves accountable for ethical development.

It's a huge collaborative undertaking.

Dr.

Reed, this has been an incredibly insightful and, at times, quite eye -opening conversation.

It really makes you realize how quickly our world is rapidly changing.

It does, JR.

Deepfake technology is a defining challenge of our digital age.

It's a powerful innovation with immense potential for good, but also a significant pathway for deception, fraud, and societal disruption.

It threatens our collective trust in digital media.

So you're saying it's not just a passing fad, but it's a fundamental shift.

Precisely.

This isn't temporary.

It's a persistent characteristic of our digital landscape, where seeing and hearing can no longer be assumed reliable.

The cat -and -mouse game between creators and detectors means a purely technological solution isn't enough.

So, how do we build a resilient digital future?

Building that future requires a multi -pronged effort.

For individuals, media literacy, critical thinking, and personal vigilance are crucial.

For tech companies, investing in robust detection, content provenance, and clear moderation policies is vital.

And for governments and policymakers, comprehensive, harmonized legislation is essential.

This includes global standards for labeling, consent, and accountability.

So, the deepfake apocalypse that wasn't yet is less a prediction and more a call to action.

It's an ongoing challenge that demands continuous adaptation and collaboration from all of us.

Exactly.

Only through a collective commitment to technological advancement, public education, and robust governance can society truly hope to navigate the complexities of synthetic media and preserve the integrity of truth in this ever -evolving digital age.

Dr.

Reed, thank you so much for joining us again on AI Innovations Unleashed.

Your insights are, as always, invaluable.

Thanks for having me.

And that's our episode for today.

So all of our listeners, thank you for tuning in to AI Innovations Unleashed, AI Unhinged, navigating the deepfake apocalypse.

So if you've liked this episode.

subscribe or like pass it along drop a comment we're out there everywhere under ai innovations unleashed so join us next week as we explore AI cults, prophets, and the quest for the singularity.

You're not going to want to miss that one.

So stay curious, my friends, stay critical, and stay safe out there in the digital wild, wild west.