Google bans Deepfake Colab Research over disinformation fears


As machine learning and artificial intelligence has evolved, deepfakes have been one of the most common and problematic uses of the tech. With the technology only getting more prominent, Google has now banned Deepfake Colab projects on its servers.

Why did Google Colab ban deepfake research?

Via Unite AI, Google is restricting deepfake projects on its Colab research servers over fears of abuse. With the technology used for anything from pornography without consent of likeness to political disinformation, Google has decided to take a stance.

Speaking to TechCrunch, a Google spokesperson explained that the company is aiming to deter deepfake abuse. With deepfake creators using the Colab platform to create fakes since 2017, there’s a responsibility there that the company can’t escape.

“Deterring abuse is an ever-evolving game, and we cannot disclose specific methods as counterparties can take advantage of the knowledge to evade detection systems," the spokesperson said. "In general, we have automated systems that detect and prohibit many types of abuse."

In recent months, deepfakes have come under even closer scrutiny over political disinformation. With fake videos of Ukrainian President Volodymyr Zelenskyy spread by Russian forces — including one that showed a fake surrender — many are asking for regulations to be made for the tech.

Of course, this isn’t the only thing that Deepfakes are used for. For example, many harmless memes and internet videos — such as Robert Downey Jr and Tom Holland in Back to the Future — are made possible by deepfakes.

Furthermore, entertainment companies are using deepfakes for the projects. For example, Lucasfilm famously used deepfakes to recreate a young Luke Skywalker in The Mandalorian. Even more recently, vocal deepfakes were used to bring back Darth Vader in Obi-Wan Kenobi.

Read More: Metaverse Sex will be better than the real thing, says Dreamcam founder

Deepfakes are not the only thing banned

While deepfakes are the more interesting story here, Google is also restricting other features on its colab platform. While the platform once allowed any Python code to run, it now restricts a large number of instances.

For example, alongside deepfakes, Colab has banned “mine cryptocurrencies, password cracking, and file hosting”. However, these bans are long-standing and are not as recent as the ban to deepfakes. 

For those who used Colab to create deepfakes, you can’t really blame Google. Any harm and abuse created on the company’s platforms is a direct result of the company’s rules, and it has a responsibility to limit that abuse.

For more articles like this, take a look at our AI page.