Data Poisoning: A ticking time bomb for Big Tech Companies

Whatever you do online, it leaves a digital footprint that serves as breadcrumbs scraped off by big tech companies who use it to track all online activities. And when we say all, we mean everything - from sending a message to streaming a movie or ordering food online. If you are ever wondering why you always get ads according to your preferences, well, this is the reason behind “personalization.”

Tech giants like Google, YouTube, Facebook, and more use these valuable data packets and feed them to their algorithms to better understand your preferences and then target you with similar ads and recommendations. This is exactly why websites like VPNRanks suggest using a VPN to make your online presence anonymous.

User data generates over $120 billion in ad revenue to Google. Sadly users don't have a say in it. This is where data poisoning comes into play.

What is Data Poisoning?

According to various sources, data poisoning is posing a real challenge to tech companies’ surveillance algorithms. In 2019, a famous reporter tried to decrease her digital footprint by cutting five major giants from her life but still failed to achieve some level of anonymity. Big data tech companies have powerful algorithms at their disposal that are capable of logging even your clicks and hovers online.

Therefore, researchers believe that the only way to redress this imbalance is by playing with data. Without the right data to mine, algorithms are useless. According to tech researchers, there are three ways users can exploit data to their advantage:

1. Data poisoning is confusing algorithms by giving meaningless data. For example, AdNauseam, a browser extension that automatically clicks on all ads served by Google, thus confusing the algorithm of your preferences.
2. Data strike is deleting or withholding your online data, so a tech firm cannot use it. It includes using privacy tools or leaving the platform altogether, so the company will have no data to use.
3. Conscious data contribution is giving meaningful data to a company that is a competitor of a platform that you are protesting again - for example, uploading your photos on Instagram instead of Tumblr.

The idea behind data poisoning is to trick algorithms by telling them stories that dont lead to anything, just like clicking on every single ad that pops up. It’s using technology to lie about your preferences, likes, and dislikes. But, you can’t change much with an individual effort. It does pose a threat to tech giants if millions of people coordinate to conduct data poisoning. For example, in January, millions of users deleted their WhatsApp accounts and moved to other messaging apps after Facebook announced that it would be using user data from WhatsApp across Facebook and other social media apps. It made Facebook delay their policy change.

Nonetheless, data poisoning seems like an effective tool for changing the present online tracking and surveillance situation. But browser extensions like AdNauseam are a great tool for confusing algorithms. The extension clicks on all advertisements shown on a webpage automatically. It shows the algorithm that the user is interested in EVERYTHING, thus sending inaccurate information which makes it harder for the system to create your profile for future use. ‘As the collected data gathered shows an omnivorous click-stream, user tracking, targeting and surveillance become futile,’ says the creator of AdNauseum extension.

Google has banned this extension from its Chrome browser, but you can install it directly from AdNauseum’s page. Therefore, data poisoning is bad news for tech giants as Google recently announced that it would stop tracking user activity across the web and also targeting ads. It is possible that the use of data poisoning tools led Google to change its policy regarding user data. However, it is hard to tell because that is something that only Google must know of, whether data poisoning is reducing the effectiveness of their algorithms or not. But banning the extension says a lot about its effectiveness.

Ali Alkhatib, a researcher at the University of San Francisco says, “It’s exciting to see this kind of work, It was really interesting to see them thinking about the collective or holistic view: we can mess with the well and make demands with that threat because it is our data and it all goes into this well together.”

It’s still too early to judge the consequences of data poisoning or to ensure that it will yield the desired results. It raises a lot of questions like how many people need to strike or poison the data to affect a company’s algorithm, or what is the most effective way? In a study involving a movie recommendation algorithm, researchers found out that if around 30% of the users went on data strikes, it affects the accuracy of the algorithm by 50%. The same simulations can be conducted on different algorithms because every machine learning system is unique, and companies regularly update their systems.

Alkhatib also says that we need more research into collective data action and how it impacts the algorithms of tech companies. But we still face a greater challenge and that is of getting millions of people to follow the data poisoning or data strike action. Similarly, there are also people who use the internet for like five seconds. How do we incorporate those people to be part of the collective action? Because this “surveillance-based model” of tech companies goes against the ethics and user’s right to privacy.

‘Google and Facebook dominate our modern lives – amassing unparalleled power over the digital world by harvesting and monetizing the personal data of billions of people.,” says Kumi Naido, secretary-general of Amnesty International. Therefore, computer scientists and researchers need to come up with more solutions and browser extensions that can feed poisoned data to the big tech companies, rendering their tracking algorithms useless and protecting user privacy.

Final Thoughts

Data poisoning is a strong tool, especially when collective action is taken against mass surveillance from tech companies. However, it’s still early to be sure about the consequences that would follow. Overall, researchers like Alkhatib are optimistic that data poisoning could change how tech companies use our data. As stated earlier, the algorithms are dependent on user data. Therefore, data leverage is a tool through which the public can turn the tables and force big companies to change their policy.