{"id":21752,"date":"2023-10-06T19:15:14","date_gmt":"2023-10-06T15:15:14","guid":{"rendered":"https:\/\/me-en.kaspersky.com\/blog\/google-privacy-sandbox-and-ad-topics-explained\/21752\/"},"modified":"2023-10-06T19:15:16","modified_gmt":"2023-10-06T15:15:16","slug":"google-privacy-sandbox-and-ad-topics-explained","status":"publish","type":"post","link":"https:\/\/me-en.kaspersky.com\/blog\/google-privacy-sandbox-and-ad-topics-explained\/21752\/","title":{"rendered":"What Google Ad Topics is, and how to disable it"},"content":{"rendered":"

Within a year, Google plans to stop supporting so-called third-party cookies<\/a><\/em>, a technology that advertisers have used for decades to track users, in its Chrome browser.<\/p>\n

But this doesn\u2019t mean such tracking will simply stop. It would be odd if the tech giant, whose revenue comes mostly from online advertising, voluntarily gave up the ability to collect user data. Instead, third-party cookies will be replaced by a new technology \u2014 Google Ad Topics. In fact, Google Ad Topics is already here: the company integrated it into the Chrome browser<\/a> this summer and recently started to roll it out to the Android operating system<\/a>.<\/p>\n

In this post, we explore how Ad Topics works, where to disable it in the Chrome and Android settings, and what else you can do to avoid being tracked<\/a> by online advertisers.\n<\/p>\n

A little history: Google Privacy Sandbox and FLoC<\/h2>\n

\nLet\u2019s first go back a bit to Google Privacy Sandbox. This is what Google calls the entire initiative to abandon third-party cookies and replace them with different technologies for targeted advertising. Google first started talking<\/a> about this initiative back in August 2019. As you can see, it\u2019s taken them four years to develop specific solutions for phasing out cookies.<\/p>\n

The purpose of this initiative is, on the one hand, to get rid of technology that\u2019s widely perceived as a violation of privacy. On the other hand, Google wants to find a way to continue showing personalized ads to users \u2014 maintaining the competitive advantage that made it an internet giant.<\/p>\n

If you look at the Wikipedia article on Privacy Sandbox<\/a>, you\u2019ll find a long list of candidate technologies that Google planned to use to move away from third-party cookies. However, in 2021, a technology called Google FLoC emerged as the primary candidate. Let\u2019s discuss it in more detail.\n<\/p>\n

What is Google FLoC?<\/h2>\n

\nFLoC (Federated Learning of Cohorts) was a technology proposed by Google aimed at changing the approach to targeted online advertising \u2014 making it more private. Instead of using individual user behavior data to personalize ads, FLoC grouped users with similar interests \u2014 meaning similar browsing histories \u2014 into \u201ccohorts\u201d. These cohorts were then assigned a unique identifier, which advertisers can use to target their ads.<\/p>\n

One of the key advantages in terms of user privacy was that FLoC didn\u2019t send user activity information to Google servers, but processed the data locally \u2014 directly on the user\u2019s device.<\/p>\n

It\u2019s worth noting that, despite the name, FLoC didn\u2019t actually use federated learning<\/em>. Its use was initially planned<\/a>, but it turned out that local computing was good enough.\n<\/p>\n

What is federated learning? It\u2019s a variant of machine learning \u2014 an alternative to the centralized learning approach. In centralized learning, data from all devices is uploaded to a central server where the AI model is trained. In federated learning, data isn\u2019t sent to a central server; instead, local models are trained on local data directly on the devices. These devices then exchange the training results, rather than the user data itself, with the server. Based on the results of this local training, a global AI model is built \u2014 thus, the local AI models and the global AI model mutually train each other. This is all done to eliminate the need for centralized storage of user data.<\/p><\/blockquote>\n

\nAs is often the case with attempts to have one\u2019s cake and eat it too, the technology was criticized from both sides. Despite Google\u2019s claims that FLoC is 95% as effective as third-party cookies, advertisers weren\u2019t satisfied<\/a> with the technology\u2019s performance.<\/p>\n

Privacy advocates, in turn, were extremely unhappy<\/a> that FLoC didn\u2019t adequately address user privacy concerns \u2014 and also that millions of Chrome users were included in the testing of FLoC without their consent. However, the main privacy concern regarding Google FLoC was its highly precise methods for categorizing users \u2014 the possible number of cohorts exceeded 30,000 \u2014 which allowed too much scope for user tracking.<\/p>\n

A significant portion of the internet industry quickly took up arms against Google FLoC, including all Chromium-based browser developers \u2014 which refused to include FLoC in their products, and Amazon \u2014 which disabled FLoC on all its websites. Consequently, just a few months after the testing of FLoC began, Google decided to freeze the project. Already by early 2022, they officially announced the abandonment of FLoC in favor of another technology \u2014 Topics API.\n<\/p>\n

What is Google Ad Topics (Topics API)<\/h2>\n

\nGoogle Ad Topics (also known as Topics API) is a technology that Google is currently planning to use to replace third-party cookies for targeted advertising. Ad Topics functions somewhat similarly to FLoC: it also relies on browser history (in the case of Chrome) or app usage (in the case of Android), through which a locally operating algorithm tries to identify user interests.<\/p>\n

However, there are significant differences: while FLoC grouped users with similar interests into cohorts with unique identifiers, Topics API only compiles a list of things that interest the user \u2014 that is, \u201ctopics\u201d.<\/p>\n

At the time of writing this article, there were 629 such topics, but this list<\/a> is constantly being updated, so the number will continue to grow. In Google Ad Topics, each webpage on the internet is associated with a certain topic. The algorithm constantly updates<\/a> the list of Google topics assigned to the user<\/em> based on what they\u2019ve visited recently. Here\u2019s how it works:\n<\/p>\n