Can De-Identification Protect Your Online Privacy?

In the last few years, the global population has an increase in privacy concerns. After the Facebook scandal in the 2016 U.S election, people start to realize how valuable their data is. But is the realization too little too late?

Most people already have either a Facebook, Twitter, or Instagram account. So has your privacy already been breached?

The answer is, probably not. Having social media accounts doesn’t automatically mean that your privacy has been breached.

By definition, privacy means the ability of an individual to seclude information of themselves. Thus allowing said individual to express themselves selectively. So, it depends on what you share on the internet.

Sharing photos on Instagram doesn’t put you at much risk other than giving data to these big tech companies. Because when the service is free you are the product. But how do I keep my data from being used against me? How do I combat the big tech overlords?

Well, one answer would be to consider using de-identification. If you are paranoid about your photos/videos being used against you. Maybe de-identifying your photos might be a solution.

What is De-Identification

De-identification or D-ID is a technique of obscuring data, making it unusable for the identification of personal identity. Data that can be de-identified are usually biometric data such as face and voice data. D-ID is increasingly being developed in the midst of the rise of artificial intelligence in modern days. Because people see D-ID as a method to protect their privacy.

I am writing about how we can use D-ID to keep our identities anonymous. The reason why I chose to write about the use of D-ID on our faces would be because of the increasing potential of mass surveillance based on what we shared on the internet.

[caption id="" align="aligncenter" width="298"] Example of De-Identification, Image retrieved from Facebook Research[/caption]

You can see an implementation of de-identification in the image above. De-identification of faces focuses on removing facial features from the subjects. As you can see some results can be more subtle than others. There are many cases in how we can use D-ID techniques to make data anonymous.

The Use Case of De-Identification

In 2014, the U.S government released a press report on Big Data and Privacy. The report says that the efficiency in technology to collect and analyze data are increasing more than people have anticipated. It would seem that even though technologies such as D-ID can be used as a safeguarded method for privacy, it might not be enough, and policies on privacy need to be enacted to do what is not technologically feasible.

Protect Sensitive Company Data

In the introduction, I mentioned some social media giants such as Facebook, Instagram, and TikTok. I mentioned them because there is an increase in the use of these platforms can have dangerous privacy risks. Especially for your facial biometric data.

Some companies that store our biometric data has a considerable risk of it being leaked. It doesn’t mean that we shouldn’t give our data when it is a requirement, it only means that the company needs to put on some more precautions regarding what and how they store their employee’s data.

Privacy in Public Datasets

The last use case I have in mind would be to protect public datasets. AI researchers need a lot of data. The best datasets are the ones retrieved from the wild. A program that automatically de-identify a person's face can benefit them anonymity. Anyone can retrieve public datasets, thus to protect subject privacy de-identification would need to be implemented before the release of the dataset.

---

Modern D-ID Solutions

Startups that offer these solutions have made a lot of heading recently. On the 26th of May 2020, a D-ID startup originating from Tel-Aviv has secured over $13.5 million of funding.

In the report, investors see the increasing performance of facial recognition as troubling. It highlights the importance of privacy protection in the 21st century. The goal of D-ID in the industry is to remove key features from a person’s face, preserving only their gender, age, and emotion. This new technology would help corporations implement procedures to protect their user data, coinciding with the new privacy laws across the globe.

Conclusion

The reason I wrote this article is to give more awareness of these kinds of technology. As modern technology enables us to do both great and terrible things. The risk of mass surveillance is real and people should be more aware of that. D-ID is a chance for us to keep our anonymity online. Safeguarding our privacy.

As of the day I write this article, there have not been any free or open source solutions that I know of. So, the only way for you to get de-identified is to implement the technique yourself by reading the research papers.