How to spot fake AI photos | Hany Farid

Summary

Podcast Summary: TED Talks Daily – How to Spot Fake AI Photos | Hany Farid

Release Date: July 2, 2025

---

Introduction

In this episode of TED Talks Daily, host Elise Hu introduces a compelling presentation by digital forensic scientist Hany Farid. Titled "How to Spot Fake AI Photos", Farid delves into the escalating dangers posed by generative AI technologies, emphasizing the transformative impact on our perception of truth and reality. This summary captures the essence of Farid’s insights, key discussions, and the subsequent Q&A session.

---

The Rising Threat of Generative AI (02:51 – 07:00)

Hany Farid opens by framing a hypothetical yet urgent scenario:

"You are a senior military officer and you've just received a chilling message on social media. Four of your soldiers have been taken. And if demands are not met in the next 10 minutes, they will be executed. All you have to go on is this grainy photo, and you don't have the time to figure out if four of your soldiers are in fact missing. What's your first move?"
[02:51]

Farid highlights his expertise in digital image authentication, developed over 30 years, which has become increasingly critical as the frequency and sophistication of manipulated images rise. He underscores the confluence of two major factors:

1. Generative AI Capabilities: The ability to create nearly indistinguishable images from reality.
2. Unregulated Social Media: Platforms amplify misinformation, making it challenging to discern truth.

"I contend that we are in a global war for truth with profound consequences for individuals, for institutions, for societies and for democracies."
[06:24]

---

Historical Context and Evolution of Image Manipulation

Farid provides a historical perspective on image manipulation, noting that while trust in photographs dates back 200 years, the ease and sophistication of altering images have dramatically increased with technological advancements:

• Victorian Era: Early forms of image manipulation for satire or political purposes.
• Digital Era: The advent of digital cameras and photo editing software made manipulation more accessible.
• Generative AI: Current technologies allow instant creation of any image imaginable, exacerbating the issue.

"From four soldiers tied up in a basement to a giraffe trying on a turtleneck sweater. It's not fun and games, of course, because generative AI is being used to supercharge past threats and create entirely new ones."
[06:22]

---

Techniques for Detecting AI-Generated Images

Farid outlines several forensic methods his team employs to authenticate digital images:

1. Residual Noise Analysis (02:51 – 04:30)

   • Fourier Transform: Examining the magnitude of noise residues to differentiate between natural and AI-generated images.
   • Indicators: Star-like patterns typically signify AI generation.
   • Quote:

     "Those star-like patterns are a telltale sign of generative AI."
     [05:45]

2. Vanishing Points and Geometry Consistency (04:31 – 05:15)

   • Vanishing Points: In natural images, parallel lines converge at a single point. AI often fails to accurately replicate this.
   • Practical Example: Railroad tracks narrowing correctly, which AI might misrepresent.
   • Quote:

     "If we can find physical and geometric anomalies, we can find evidence of manipulation or generation."
     [06:05]

3. Shadow Analysis (05:16 – 06:00)

   • Shadow Consistency: Natural images have shadows that adhere to the laws of physics. AI-generated images often exhibit unrealistic shadow behavior.
   • Quote:

     "Shadows... give a very good indication that this image is not authentic."
     [06:00]

Farid emphasizes that no single forensic technique is infallible, advocating for a multi-faceted approach to authentication.

---

Societal Implications and the Battle for Truth (06:01 – 09:50)

Farid articulates the broader societal impacts of fake AI-generated content:

• Erosion of Trust: Difficulty in believing any online content leads to widespread skepticism.
• Use Cases of Malicious Intent:
  • Extortion: Creating fake compromising images to humiliate individuals.
  • Misinformation: Fake videos of professionals (e.g., doctors promoting false cures).
  • Corporate Fraud: Impersonating executives to deceive organizations, resulting in significant financial losses.

"We are all vulnerable. It's useful to understand how generative AI works..."
[06:30]

Farid warns of a "global war for truth," highlighting the dire consequences for democracies and societal cohesion if trust in digital content continues to deteriorate.

---

Solutions and Recommendations (09:51 – 12:50)

Hany Farid proposes several strategies to combat the spread of fake AI-generated images:

1. Advanced Forensic Tools (09:51 – 10:30)

   • Development and accessibility of sophisticated tools for journalists, courts, and institutions to verify authenticity.
   • Quote:

     "The tools that I've described... are being made available to journalists, to institutions, to the court to help them tell what's real and fake."
     [11:45]

2. International Content Credentials (10:31 – 11:15)

   • Establishing standards that authenticate content at the point of creation.
   • These credentials will assist consumers in identifying genuine content.
   • Quote:

     "As these credentials start to roll out, they will help you, the consumer, figure out what is real and what is fake online."
     [11:00]

3. Caution with Social Media Usage (11:16 – 11:55)

   • Advising against using social media as a primary news source due to its prevalence of misinformation.
   • Encouraging reduced reliance on platforms that prioritize engagement over accuracy.
   • Quote:

     "Please understand that social media is not a place to get news and information... it is simply too riddled with lies and conspiracies."
     [11:35]

4. Personal Responsibility in Information Sharing (11:56 – 12:50)

   • Urging individuals to verify information before sharing to prevent the perpetuation of falsehoods.
   • Recognizing the collective responsibility in maintaining the integrity of the online information ecosystem.
   • Quote:

     "Take a breath before you share information and don't deceive your friends and your families and your colleagues."
     [12:20]

Farid closes with a pivotal choice facing society:

"We're at a fork in the road. One path... allows technology to rip us apart... The other path... leverage the power of technology to work for us and with us."
[12:45]

---

Q&A Session with Latif Nasser (12:54 – 13:50)

Following his talk, Farid engages in a brief Q&A with Latif Nasser, co-host of Radiolab.

1. Percentage of Fake Images Online

   • Farid estimates that nearly 50% of images online might be fake, varying by platform.
   • Quote:

     "I would say we're getting close to 50%."
     [13:01]

2. Differentiating Filtered vs. AI-Generated Images

   • While detection is possible, the task is becoming increasingly challenging.
   • Quote:

     "Yes, but it's becoming increasingly more difficult."
     [13:19]

3. Tools for Laypersons to Verify Images

   • Farid suggests that as more fake content is created, authentication is becoming complex, cautioning against excessive reliance on dubious verification websites.
   • Quote:

     "Don't do it."
     [13:26]

4. Capability to Enhance Images Like in CSI Shows

   • Farid confirms the ability to enhance images, aligning with forensic practices.
   • Quote:

     "Yes."
     [13:43]

---

Conclusion

Hany Farid's enlightening talk underscores the critical challenges posed by generative AI in the realm of digital authenticity. By elucidating the techniques for detecting fake images and emphasizing the societal implications, Farid calls for a collective effort to preserve truth in the digital age. His recommendations advocate for technological advancements, regulatory standards, and personal responsibility to navigate the complex landscape of information reliability.

---

Notable Quotes:

• "We are in a global war for truth with profound consequences for individuals, for institutions, for societies and for democracies."
  — Hany Farid ([06:24])

• "Those star-like patterns are a telltale sign of generative AI."
  — Hany Farid ([05:45])

• "Please understand that social media is not a place to get news and information... it is simply too riddled with lies and conspiracies."
  — Hany Farid ([11:35])

• "We're at a fork in the road... The other path... leverage the power of technology to work for us and with us."
  — Hany Farid ([12:45])

---

This episode serves as a crucial wake-up call, urging listeners to critically evaluate the authenticity of digital content and to take active measures in safeguarding the integrity of information dissemination in the era of AI.

Transcript

[Elise Hu]

You're listening to TED Talks Daily where we bring you new ideas and conversations to spark your curiosity every day. I'm your host Elise Hu're looking at a grainy photograph of soldiers who've been taken hostage. Is this photo real or fake? Until recently, this question wouldn't have been difficult to answer, but today it may be the first thing we need to ask. In this talk, digital forensic scientist Han Hani Fareed warns of the fast approaching dangers of generative AI in forever changing our understanding of truth and facts, and says that when it comes to our engagement with technology, we're at a pivotal fork in the road. It all comes down to what choice we will make. Stick around after his talk for a brief Q and A between Hani and Latif Nasser, the co host of Radiolab and a guest curator, ed2025 and tune.

[Georgie Frost]

Into this very feed later today for.

[Elise Hu]

A special conversation between Hani Farid and me where we dig into some of the deeper ideas from his talk.

[Georgie Frost]

Want to go deeper on the business topics that matter? This is the so what from BCG and I'm your host Georgie Frost. BCG experts explore the most pressing topics on the minds of the world's top business leaders. Listen wherever you get your podcasts.

This message is brought to you by Apple Card. Each Apple product, like the iPhone 16, is thoughtfully designed by skilled designers. The titanium Apple Card is no different. It's laser etched, has no numbers and it earns you daily cash on everything you buy, including 3% back on everything at Apple. Apply for Apple Card on your iPhone in minutes, subject to credit approval. Apple Card is issued by Goldman Sachs Bank USA Salt Lake City Branch terms and [e-mail] this episode is sponsored by Dell introducing the new Dell AI PC. Powered by the Intel Core Ultra processor. It's not just an AI computer, it's a computer built for AI. That means it's built to help do your busy work for you so you can fast forward through editing images, designing presentations, generating code, debugging code, running lots of apps without lag, creating live translations and captions, summarizing meeting notes, extending battery life, enhancing security, finding that file you are looking for, managing your schedule, meeting your deadlines, responding to Jim's long emails, leaving all the time in the world for more you time and for the things you actually want to do. No offense Jim. Get a new Dell AI PC starting at [e-mail] AI PC how those ahead? Stay ahead.

[Hani Hani Fareed]

You are a senior military officer and you've just received a chilling message on social media. Four of your soldiers have been taken. And if demands are not met in the next 10 minutes, they will be executed. All you have to go on is this grainy photo, and you don't have the time to figure out if four of your soldiers are in fact missing. What's your first move, if I may be so bold? Your first move is to contact somebody like me and my team. I am by training an applied mathematician and computer scientist. And I know that seems like a very strange first call at a moment like this, but I've spent the last 30 years developing technologies to analyze and authenticate digital images in digital videos. Along the way, we've worked with journalists, we've worked with courts, we've worked with governments on a range of cases, from a damning photo of a cheating spouse, gut wrenching images of child abuse, photographic evidence in a capital murder case, and of course, things that we just can't talk about. Used to be a case would come across my desk once a month, and then it was once a week. Now it's almost every day. And the reason for this escalation is a combination of things. One, generative AI. We now have the ability to create images that are almost indistinguishable from reality. Two, social media dominates the world and is largely unregulated and actively promotes and amplifies lies and conspiracies over the truth. And collectively, this means that it is becoming harder and harder to believe anything that we read, see or hear online. I contend that we are in a global war for truth with profound consequences for individuals, for institutions, for societies and for democracies. And I'd like to spend a little time talking today about what my team and I are doing to try to return some of that trust to our online world and in turn, our offline world. For 200 years, it seemed reasonable to trust photographs. But Even in the mid-1800s, turns out the Victorians had a sense of humor. They manipulated images. Or you could alter history. If you fell out of favor with Stalin, for example, you may be airbrushed out of the history books. But then in the turn of the millennium, with the rise of digital cameras and photo editing software, it became easier and easier to manipulate reality. And now with generative AI, anybody can create any image of anything anywhere at a touch of a button. From four soldiers tied up in a basement to a giraffe trying on a turtleneck sweater. It's not fun and games, of course, because generative AI is being used to supercharge past threats and create entirely new ones. The creation of nudes of real women and children used to humiliate or extort them. Fake videos of doctors promoting bogus cures for serious illnesses. A Fortune 500 company losing tens of millions of dollars because an AI impersonator of their CEO infiltrated a video call. Those threats are real. They are here. And we are all vulnerable. It's useful to understand how generative AI works, starting with billions of images with a descriptive caption. Each image is degraded until nothing but visual noise is left, a random array of pixels. And then the AI model learns how to reverse that process by essentially turning that noise back into the original image. And when this process is done, not once, not twice, but billions of times on a diverse set of images, the machine has learned how to convert noise into an image that is semantically consistent with anything you type. And it's incredible, but it is decidedly not how a natural photograph is taken, which is the result of converting light that strikes an electronic sensor into a digital representation. And so one of the first things we like to look at is whether the residual noise in an image looks more like a natural image or an AI generated image. Those star like patterns are a telltale sign of generative AI. Now, for the mathematicians and the physicists in the audience, that is the magnitude of the Fourier transform of the noise residual. For everybody else, that detail doesn't matter. But you definitely should have taken more math in college. Professors can't help themselves, but no forensic technique is perfect. And so you don't stop after one thing. You keep going. So let's go on to our next one, the vanishing points. If you image parallel lines in the physical world, they will converge to a single point, what's called a vanishing point. Good intuition for that is the railroad tracks. Railroad tracks are obviously parallel. They narrow as they recede away from me and intersect at a single vanishing point. This is a phenomenon that artists have known for centuries. But here's the great AI doesn't know this, because AI is fundamentally, as I just described, a statistical process. It doesn't understand the physical world, the geometry, and the physics. So if we can find physical and geometric anomalies, we can find evidence of manipulation or generation. Evidence number two. All right, what else can we learn? Surprisingly, shadows have a lot in common with vanishing points. And again, this is a physical phenomena that you expect in natural images. And because AI fundamentally doesn't model the physics and the geometry of the world, and it tends to violate these physics, we now have a very good indication that this image is not authentic. The most important thing I want you to take away from this is that while it may not be easy, it is possible to distinguish what is real from what is fake. I think this image is a bit of a metaphor for how a lot of us feel. We feel like hostages. We don't know what to trust anymore. We don't know what is real, what is fake. But we don't have to be hostages. We don't have to succumb to the worst human instincts that pollute our online communities. We have agency and we can affect change. Now, I can't turn you all into digital forensics experts in 10 minutes, but I can leave you with a few thoughts. 1. Take comfort in knowing that the tools that I've described and that my team and I are developing are being made available to journalists, to institutions, to the court to help them tell what's real and fake, which in turn helps you. 2. There is an international standard for so called content credentials that can authenticate content at the point of creation. As these credentials start to roll out, they will help you, the consumer, figure out what is real and what is fake online. And while they won't solve all of our problems, they will absolutely be part of a larger solution. 3. Please understand that social media is not a place to get news and information. It is a place that Silicon Valley created to steal your time, your attention by delivering you the equivalent of junk food. And like, thank you. And like any bad habit, you should quit. And if you can't quit, at least do not let this be your primary source of information because it is simply too riddled with lies and conspiracies and now AI slop to be even close to being reliable. 4. Understand that when you share false or misleading information, intentionally or not, you're all part of the problem. Don't be part of the problem. There are serious, smart, hardworking journalists and fact checkers out there who work every day because I talk to them every day to sort out the lies from the truths. Take a breath before you share information and don't deceive your friends and your families and your colleagues and further pollute the online information ecosystem. We're at a fork in the road. One path. We can keep doing what we've been doing for 20 years, allowing technology to rip us apart as a society, sowing distrust, hate, intolerance. Or we can change paths. We can find a new way to leverage the power of technology to work for us and with us and not against us. That choice is entirely Ours. Thank you.

[Latif Nasser]

We're doing rapid fire questions. You ready? Roughly what percent of images online do you believe to be fake?

[Hani Hani Fareed]

Depends on the platform. Signal to noise ratio is getting close to one. Stay off of Twitter, stay off of X, and stay off of everything else for that matter.

[Latif Nasser]

But give me. So what do you.

[Hani Hani Fareed]

I would say we're getting close to 50%. Wow.

[Latif Nasser]

Can you differentiate between things that have like Instagram or TikTok filters or Photoshop versus fully AI generated images?

[Hani Hani Fareed]

Yes, but it's becoming increasingly more difficult.

[Unknown]

Huh.

[Latif Nasser]

Are there any websites a layperson can use to check?

[Hani Hani Fareed]

Whoa. Okay. By the way, this is a secondary problem, which is now people are creating fake things than going to fake sites to authenticate fake things. And it's all getting very weird. Don't do it.

[Latif Nasser]

Okay, last one, most important one. In CSI crime shows, when they say enhance. Can you do that?

[Hani Hani Fareed]

Yes.

[Latif Nasser]

Okay, great. Hanif, everybody.

[Elise Hu]

That was Hani for Reid speaking at TED 2025. If you're curious about TED's curation, find out [e-mail] curationguidelines and that's it for today's show. TED Talks Daily is part of the TED Audio Collective. This episode was produced and edited by our team, Martha Estefanos, Oliver Friedman, Brian Greene, Lucy Little, Alejandra Salazar and Tonsika Sarmarnivon. It was mixed by Christopher Faizy Bogan. Additional support from Emma Tob and Daniela Ballaraizo. I'm Elise Hu. I'll be back tomorrow with a fresh idea for your feed. Thanks for listening.

[Georgie Frost]

This episode is brought to you by Progressive Insurance. Do you ever think about switching insurance companies to see if you could save some cash? Progressive makes it easy. Just drop in some details about yourself and see if you're eligible to save money when you bundle your home and auto policies. The process only takes minutes and it could mean hundreds more in your pocket. Visit progressive.com after this episode to see if you could save Progressive Casualty Insurance company and affiliates. Potential savings will vary. Not available in all states.

[Unknown]

Listen up. You can get the new iPhone 16e with Apple Intelligence for just 49.99 when you switch to Boost Mobile. We pulled so many all nighters to give you this deal and hey, stop messing with the mic.

[Georgie Frost]

I'm just helping this catch people's attention.

[Hani Hani Fareed]

This is a great.

[Unknown]

Exactly. So it doesn't need all that.

[Hani Hani Fareed]

Fine.

[Unknown]

Get the new iPhone 16e available at Apple Store locations and the Apple Store online.

[Latif Nasser]

Visit your nearest Boost Mobile store for full offer details. Apple Intelligence requires iOS 18.1 or later restrictions apply.

[Unknown]

If you work as a manufacturing facilities engineer, installing a new piece of equipment can be as complex as the machinery itself. From prep work to alignment and testing, it's your team's job to put it all together. That's why it's good to have Grainger on your side. With industrial grade products and next day delivery, Granger helps ensure you have everything you need close at hand through every step of the installation. Call 1-800-granger clickgranger.com or just stop by Granger for the ones who get it done.