Expert calls for tougher regulation as deepfake threats spiral

Fake news and fake nudes have taken over the internet, and still, nobody seems to be doing anything about it.

Deepfake technology, particularly nudify apps and chatbots, has received nothing but bad press.

The most notable scandal involved Elon Musk’s chatbot Grok. Its rebellious and “anti-woke” approach allowed users worldwide to exploit women and, reportedly, children through deepfake pornography.

While the media is full of reports that vilify deepfakes, there could be some positive use cases for this technology.

Cybernews spoke to Dr. Manny Ahmed, the Cambridge-educated founder and CEO of OpenOrigins, a tech company that claims to “build global trust back into visual content,” about the implications of deepfake content, and what should be done to reduce harm.

Deepfakes aren’t all bad, right?

Ahmed’s company claims that deepfakes “threaten our shared understanding of reality,” and the startup is focusing on showing users what’s real instead of exclusively identifying artificially generated content.

With this in mind, I wanted to see if Ahmed could tell me about positive use cases for deepfake technology to gain a more nuanced understanding.

There are innocuous uses for deepfakes, like translating company training content into different languages and creating avatars for visual learning, Ahmed told us.

However, intent and context are arguably the most important factors to consider when looking at deepfake content, particularly when it’s manipulating an image of a real person.

On the surface, “the harm that comes from creating a deepfake is minimal,” as, at the end of the day, it’s just a tool.

The versatility of the tool doesn’t make it less explosive

While deepfake technology itself is just a tool, Ahmed conflates it with materials like gunpowder.

“Gunpowder can be used to make bullets…but can also be used to make roads,” Ahmed said.

Fun fact, gunpowder was discovered by accident in China when a Taoist alchemist mixed three chemical components in the hope of discovering the elixir of eternal life.

Instead, the ancient chemist created what some consider one of the deadliest and most significant inventions in human history, besides the atomic bomb.

I was surprised by the versatility of gunpowder and naively believed that the black powder itself could be emulsified and turned into bricks or used to pave roads.

But I soon became aware that gunpowder is used solely for its explosive properties. This comparison got me thinking: Is deepfake technology naturally hazardous?

If gunpowder has the propensity to explode and the by-products of this reaction are harmful to living organisms, surely that makes it bad, right?

Is the tool or the user to blame when deepfakes do harm?

It’s true that it’s difficult to murder someone with a spoon. While not impossible, it’s an ineffective tool for the job.

But, if a spoon is used to murder someone, does it make the spoon itself bad?

The argument here is that tools themselves aren’t inherently good or bad. It’s just “easier to misuse” certain things.

For example, a user could use deepfake technology to “create a pornographic video of someone without their consent and publish it…which could ruin their reputation.”

Ahmed seems to imply that deepfake porn generation is the most common use case and positive applications of deepfake technology only “happen in such secluded enterprise use cases.”

“It’s very hard to kill someone with a rubber bullet”

Chatbots like Musk’s Grok have arguably democratized the misuse of these tools.

Take bad actors hacking devices, there’s a level of sophistication and technical know-how required to carry out such an attack, Ahmed explained.

“Whereas deepfakes, you just ask Grok to make a pornographic video for you. So it democratized the negative use cases.”

When bad actors are given tools that lower the barrier for entry into cybercrime, particularly tools like Grok, which are designed with limited guardrails, you’re asking for trouble.

It’s like if you give a murderer a choice of two guns, one with steel bullets and the other with rubber bullets.

“It’s very hard to kill someone with a rubber bullet…it’s very easy to kill someone with gunpowder or with an actual bullet.”

This is one of the many reasons Ahmed argues that deepfake technology should be regulated.

If governments and big tech companies do more to cushion the blow, they could then equip malicious users with rubber bullets instead of the real thing.

“I would argue that this is one of those technologies that needs to be very heavily regulated,” Ahmed said outrightly.

There’s a reason why there are gun regulations in countries that allow their civilians to carry arms.

While guns are still misused, there are regulations in place to ensure that the majority of people who carry arms don’t abuse them.

“That’s why we don’t allow people to buy and create gunpowder to do whatever they want with it – we regulate it,” Ahmed said.

The same should be said for deepfake technology, yet “regulators are sleeping at the wheel.”

“They’ve done some bare minimum regulations. The European Union is passing a law that makes it illegal to create non-consensual pornography using AI. Great. We should definitely do that, but we need to go a step further.”

I asked Ahmed what the best course of action would be, coming from someone in the industry.

In essence, governments need to regulate the companies that produce these models.

OpenAI, xAI, Google, Anthropic, and all the others.

“There are only basically five companies in the world that make all of these deepfakes, so, regulate them…tell them that you’re not allowed to produce these kinds of outputs. They have the tools to monitor what's actually happening.”