Google resumes shoveling stuff into its 'Privacy Sandbox'

Chrome trials scheduled for FLEDGE and Topics APIs


Google is preparing another round of tests for the latest iteration of its purportedly private-preserving ad technology, after last year's Federated Learning of Cohorts (FLoC) experiment revealed the need for further refinement.

In separate messages to Chromium developers declaring their "Intent to Experiment," Google software developers on Friday said Origin Trials for the company's FLEDGE API and its Topics API will commence following the March 31 debut of Chrome 101 Beta. Testing is expected to continue at least until Chrome 104 Beta, three months hence.

FLEDGE aims to enable remarketing – showing ads at a website based on prior interactions at a different website – and Topics, which replaces FLoC, aims to enable interest-based advertising. And both aspire to do so in a way that doesn't involve tracking individuals across the web, or so it's said.

FLEDGE is an effort to implement Turtledove, an API to facilitate advertising targeted at interest groups. It moves interest data and the decision about which ad gets presented from the server-side to the client-side (browser), for the sake of privacy.

"The intent of the Topics API is to provide callers (including third-party ad-tech or advertising providers on the page that run script) with coarse-grained advertising topics that the page visitor might currently be interested in," Google says.

Google's explainer for FLEDGE/Turtledove says that the in-browser ad scheme "involves the browser running untrusted JavaScript downloaded from multiple parties" and outlines various ways in which the API imposes limitations on the execution environment for the sake of security.

The Topics API also has security and privacy considerations that have yet to be fully addressed.

Google's hope in experimenting with these APIs is to prove that FLEDGE and Topics are privacy-preserving and revenue-preserving, as well as secure.

Since the early days of web advertising, presenting ads to people using web browsers has involved cookies – files that get deposited by web server code on behalf of the site publisher and affiliated third-party firms.

As the privacy problems posed by this approach became apparent and spurred regulation, and as Google's competitors made changes to restrict the use of third-party cookies, Google in 2019 launched its Privacy Sandbox initiative to redesign its ad tech in a way that complies with evolving privacy rules and tolerates privacy defenses.

With that project underway, Google in January 2020 announced plans to phase out third-party cookies "within two years," a commitment soon thereafter hedged with qualifiers. By the middle of last year, the third-party cookie phase out had slipped back to late 2023.

A matter of trust

Part of the problem for Google is that ad industry rivals fear they will be at a data disadvantage in the Privacy Sandbox and their concerns have reached the ears of lawmakers and regulators in the US, Europe, and the UK at a time when the ad biz faces broad antitrust scrutiny and litigation.

The result has been that Google made a set of commitments to the UK's Competition & Markets Authority that it will design its Privacy Sandbox systems in consultation with competitors. So now instead of moving fast and breaking things, the online ad giant has to engage with marketers who think this whole privacy push will put them at a disadvantage.

Google also faces ongoing criticism from rival browser makers like Brave that argue its Privacy Sandbox only improves privacy as measured from the intrusive baseline set by Chrome. The Topics API, said Brave senior director of privacy Peter Snyder in January, is dangerous because it makes Google the arbiter of what data is "sensitive" in terms of the interests associated with a particular internet user.

And FLEDGE, Snyder has cautioned, relies on WebBundles, which pack web resources for download. They pose a security and privacy threat, he contends, because they remove resources from the global namespace, where they can be identified and blocked. Content blocking extensions would not be able to block bundled resources because they would not know what file name or string to look for.

"Anyone with a concern for a truly privacy-first Web should be concerned with Fledge and Topics API," said Snyder in an email to The Register. "Google is trying to track the web on a course that still favors their infrastructure and advantages, before others can nudge things on a more user-focused approach."

"Much of 'Privacy Sandbox' should be understood as 'moat around Google,' where Google is pushing for a direct or intermediating role in an ever greater percentage of web requests, knowing they have first-party access to nearly every site (Google Analytics, AdWords, Google Tag Manager, Google Maps, etc)." ®


Other stories you might like

Biting the hand that feeds IT © 1998–2022