Protecting Our Children Against the Rise of Deepfake AI Exploitation

Personal thoughts on a Spanish Community’s struggle against the dark side of AI technology and synthetic media

I’ve written about some scary stuff about AI and synthetic media, from deathbots to high-end AI girlfriends, but nothing scares me more than deepfakes.

If chatbots have a limited scope, deepfake technology can be used in an almost limitless range and that alone is enough to give me the heebie-jeebies.

The deep-learning algorithms behind it can emulate an original image and virtually twist it in any way you can imagine and that’s what I call scary.

In the wrong hands “synthetic content” is then easily exploited for the most devious and malignant purposes like revenge porn, extortion scams, identity theft, or simply for spreading misinformation.

For a little bit of context we have to go back a decade. Deepfakes came to the limelight in the 2010s when a Reddit user under the moniker “deepfake” created a computational graphics algorithm and used it to swap faces and begin sharing videos of celebrities in fake sex scenes.

A lot has happened in the world of AI since then and nowadays you don’t need to be a computacional ninja with a wide set of IT skills to create deepfakes. Any high school bully armed with a smartphone can do it.

That probably explains why every other day, I read a disturbing story about the alarming rise of artificial intelligence in the sexual coercion and exploitation.

Hence, it makes me think we cannot ignore the fact that deepfakes have the potential to mess up people’s lives and drive them close to the abyss.

I know I’m not alone, as many of us share this concern about the role of AI and deepfake technology.

Recent events in neighboring Spain have made me think seriously about this issue. The account is told in the first person by a mother who witnessed her daughter fall victim to the disturbing dark side of AI technology.

Sofia (I’ve altered all the names) couldn’t believe what was happening in their small town of Almendralejo, Badajoz.

The mother took it to Instagram, where she has thousands of followers. In the video, she explains how her daughter, Isabella, came to her sobbing:

“Mom, look at what they’ve done to me; they’ve taken one of my photos and made it look like I’m naked.”

Sofia looks in awe at the photo her daughter holds on her phone, and she thinks to herself:

“Anyone who sees that thinks it’s real.”

She adds:

“I know my daughter, but anyone who sees that thinks it’s real.”

That’s the evil side of innovation, AI has evolved to the point that now you can’t say what is real and what is not.

The mother goes on to explain how the people who did this to dozens of girls from their town are other teenagers from the community:

“This video goes to those who have done this barbarity, to those teenage boys. You are not aware of the damage you have done and the crime that this is. It is a crime and a serious one.”

Sofia is right. She knows that when you upload anything to the internet, it never goes away. The fake images may have already found their way onto the Dark Web, and then it becomes almost impossible to make them disappear.

The girls who have fallen victim to this must understand one thing, Ximena adds:

“You are not to blame for what has happened.”

Almendralejo is a small community in the interior of the country with only around thirty thousand inhabitants, and the news of what is happening is whispered by everyone, but only a few will speak openly about it. Those who seem deeply troubled to see this happening to children.

When asked, they will emphasize how many of the girls were only 12–14 years old. The mothers were quick to report the case to the authorities, and the Spanish National Police is investigating the dissemination of nude images of minors.

The Spanish Police became aware of the first cases in July. According to the authorities, there should be more than thirty girls from several schools around town. Seven of them have already made statements.

An official statement from the government of Extremadura to the media on Monday explains how the suspects are other teenagers who have resorted to an application that generates fake nude images using artificial intelligence from photos of the girls easily found on social media.

The Spanish Juvenile Prosecutor’s Office has asked the matter to be dealt with with “discretion” because apparently “all of those involved in it, both the victims and the possible perpetrators, are minors”.

One of the Almendralejo victims shared with the police an exchange with a boy in which he urged her to pay him money, and when she refused, the boy sent a doctored nude image of her.

The latest news coming from Spain on this topic goes on to state how The Government of Spain considers it “urgent” to approve the new regulation on artificial intelligence (AI). This is nothing new, I’ve heard it many times before. In Spain, as here in Portugal, we talk too much and do too little.

Everyone is calmly waiting for the European Union to come up with new regulations to address the risks of this technology, but little is done in the field to prevent it.

Reading an article about how one of the top go-to representatives in the country’s Digital Transformation policies believes a possible solution is to have “a watermark” on images that have been manipulated with the use of artificial intelligence is preposterous.

How does this help the victims?

I agree that European Institutions have been at the forefront of artificial intelligence regulation. The EU is one of the first jurisdictions in the world with a legal regime on AI that is increasingly being used as a global standard. But laws are just words on paper if we don’t find ways to better enforce them.

In 2019, UNICEF released data from a poll showing how “young people in 30 countries said they have been victims of online bullying, with one in five reporting having skipped school due to cyberbullying and violence.”

Cyber exploitation is on the rise. “In 2020, it was estimated that an AI bot operating on Telegram generated 100,000 pornographic ‘deepfakes’ of real women and girls. These figures are more than alarming and should raise widespread societal alarm.

Thus, sextortion (cybersexual exploitation) in the AI era is a major concern that needs attention.

Here are some key points to consider:

1. Extortion and exploitation: Extortion schemes often involve the creation of content using deepfake technology. These scams often target women and children, who may face threats of exposing fabricated content unless they submit to the aggressor's request.
2. Impact on individuals: cyberbullying and exploitation cause humiliation, fear, anxiety, and even physical harm. Victims will suffer distress and face detrimental effects on their professional lives and overall well-being. When we’re talking about young children, the impact is undeniably overwhelming.
3. Consent and privacy violations: Deepfake pornography is regarded as a form of violence that violates consent rights. It involves the usage of someone's likeness without their permission. Harms both the victim and the original adult performer involved.
4. Insufficient legal framework: Currently, there is a lack of measures to address deepfake exploitation issues globally.

A few U.S. states have laws tackling this problem. However, several studies emphasize how the country lacks federal legislation to address the potential threats of deepfake technology, even if the National Defense Authorization Act (NDAA) from 2020 onwards has, and I quote, “included provisions that address the growing problem of deepfakes.”

The National Security Agency (NSA), the Federal Bureau of Investigation (FBI), and the Cybersecurity and Infrastructure Security Agency (CISA) recently released a Cybersecurity Information Sheet (CSI), Contextualizing Deepfake Threats to Organizations, hence, they are addressing the issue of synthetic media threats. Even local police authorities are becoming more aware of it.

Unfortunately, this is not enough.

It’s crucial for society as a whole to recognize these challenges posed by deepfake exploitation in order to develop solutions that protect individuals’ rights and mitigate harm caused by this misuse of AI technology.

Companies on the vanguard of Gen-AI could play a role in developing new ways to help prevent cyber-exploitation instead of simply striving for innovation that generates gargantuan profits for their stakeholders.

Closing thoughts

I agree that governments and authorities have to enforce measures that limit access to harmful content and how it “normalizes” inappropriate behaviors or stereotypes.

Still, I don’t think the way forward lies merely in a technological solution. Using AI to detect doctored or altered images circulated online can help mitigate the problem, but as in the Almendralejo case, when circulated among peers, deepfakes have already left an undeletable mark on the victims long before they are widely spread online.

So the solution can only be found upstream; we have to address the root of the problem, namely when we are talking about cyberbullying and sexual coercion happening among peers at a young age.

Hence, we can do our part by promoting values such as equality and respect for others. You can lobby against cyberbullying and exploitation of any kind. And help raise awareness regarding the trivialization of human sexuality and the lack of adequate affective-sexual education.

The real problem is not whether machines think but whether men do — B. F. Skinner