Facebook is attempting to develop the next generation of targeted ad systems that protect individual user data. Privacy-Enhancing Technologies or PET relies on cryptographic and statistical techniques for ad personalization.
Facebook is building a new AdTech platform that promises to offer ad-measurement and personalization while minimizing data collection. The company is calling it Privacy-Enhancing Technologies or PET, and it may sound similar to Google’s Federated Learning of Cohorts (FLoC). However, there are significant differences.
Facebook’s data collection methods facing an uncertain future with Apple’s ATT and Google FLoC:
Facebook is one of the largest collectors of user data. Privacy advocates have routinely suggested abandoning the social media platform to anyone who values their privacy.
Recent changes are proving quite detrimental to Facebook and more specifically, to its data collection practices. Apple’s App Tracking Transparency (ATT) is already one of the biggest hurdles that are proving to be exceptionally difficult to overcome.
Facebook to rebuild its ad business with privacy-enhancing technologies (PETs) to minimize data collection and improve user safety. https://t.co/eAUbMcFR3v#Facebook #PETs #digitaladvertising #advertising
— Mobile Marketing Reads (@mmarketingreads) August 12, 2021
Apple recently offered a system-level prompt that asks iPhone and iPad users if they want to grant their consent to tracking. More than 90 percent of U.S. residents have reportedly denied their consent. And Apple Inc. is receiving a similar response across the world.
While there’s Apple’s ATT on one side, there’s Google FLoC on the other. The search giant may have delayed the deployment of FLoC, but there are no plans to scrap the same.
Today we're sharing more details about how we are approaching the next era of personalized experiences and building a portfolio of privacy-enhancing technologies: https://t.co/pytSBH67IZ
— Alexandru Voica 💀 (@alexvoica) August 11, 2021
Technical jargon aside, FLoC promises to herd users with similar interests into flocks and anonymize the user data. Advertising companies will have access to such aggregated data, and there won’t be any data that identifies individuals.
Facebook’s new PET may appear similar to Google’s FLoC. However, the social media giant is relying on multiple technologies to further protect user privacy and user data. The company will still serve personalized advertising, but advertisers won’t have access to data on individuals.
Facebook revamping its advertising system to prioritize user privacy and reduce data collection with Privacy-Enhancing Technologies:
Facebook has revealed more details about how it plans to use Privacy-Enhancing Technologies (PETs) to power the next generation of digital advertising.
The social media giant claims it is using techniques based on cryptography and statistics to implement PETs. These supposedly reduce the data that Facebook must collect and process.
Are you ready 👀 for Privacy Enhancing Technologies?
But how do these worlds connect? 👇 https://t.co/lobYrcqBVl
— Ericsson (@ericsson) August 12, 2021
Even with the reduced amount of user data, Facebook claims it can offer ad accuracy and personalized preferences. The company is using technologies such as secure Multi-Party Computation (MPC), on-device Machine Learning, and Differential Privacy.
Additionally, Facebook began testing a system called Private Lift Measurement in 2020. The system allows advertisers to see how campaigns are performing while limiting what user data is available to them or Facebook itself.
There are some good explainers out there for on-device learning and differential privacy that are widely accessible to a broad audience. I haven’t seen similar for Secure Multiparty Computation (MPC). Our blogpost about PETs touches on this topic briefly: https://t.co/8pTqIdpVQt
— Ben Savage (@BenSava78155446) August 11, 2021
The social media giant has also open-sourced a platform that allows third parties to develop more private ad measurement products. One of the major technologies, however, will undoubtedly be on-device Machine Learning.
Incidentally, Apple Inc. is widely using on-device learning. This significantly reduces data exfiltration while still allowing highly targeted ads.