Google's next big idea for browser security looks like another freedom grab to some
Safe to say, this proposal has gone down like a
Analysis Googlers have proposed a way to determine whether browsers can be trusted, as a defense against criminal fraud and other bad behavior. Some in the internet community fear this is the end of the web as we know it.
The proposal, dubbed Web Environment Integrity (WEI), showed up as code in April and was announced in May. It elicited a handful of concerned comments among those who follow the development of the Chromium open source project's Blink rendering engine, but didn't attract much attention from the technical community until it was published on Friday as a working draft specification.
Google's engineers describe WEI as a way for browser clients to establish trust with a server through a third party (eg, Google Play) that presents a token attesting to the integrity of the client environment.
In simpler terms, WEI provides a way for a browser to prove it is working as a website operator expects, and hasn't been manipulated. If you have a website that offers in-browser gaming, and you want to make sure no player is cheating, you could use WEI to determine that connected clients are pure, legit, and not running any cheat code.
Same goes for websites that don't want automated bots posting or liking posts – engagement has to be done via an accepted, unaltered browser. And for publishers that only want to serve content and ads to browsers that definitely aren't just bots.
This therefore starts to slide the web toward a time in which only authorized, officially released browsers will be accepted by websites.
And since Chromium serves as the foundation of not just Google Chrome, but also Microsoft Edge, Brave, and a number of other browsers, WEI could have a broad effect on the web – if and when it gets deployed and adopted.
"The Web Environment Integrity API allows user agents to request attester verdicts from an attester that can be used to verify the integrity of the web environment," the draft spec explains. "These verdicts are piped to a relying party where they are validated for authenticity. Web Environment Integrity is best suited for detecting deceptive web environments."
The proposal's lack of detail at this stage is evident in the link that explains "web environments" as a todo item.
The stated purpose of the API is to address various long-standing problems on the web: social media manipulation and fakery; bot detection; misuse of WebViews in apps; bulk web hijacking and account creation; cheating in web-based games; compromised devices; and password-guessing attempts.
However, "abuse" is not specifically defined. So when the authors of the spec say a goal is to "offer an adversarially robust and long-term sustainable anti-abuse solution," it's not clear what would be disallowed.
Same old, same old
The idea – bringing trust to web interactions – is not new. Similar APIs for validating native apps in the Android and iOS ecosystems already exist. There are proposals with related aims – like PrivacyPass, the Trust Token API, and UserConfidenceScore. A precursor to WEI was initially proposed in April, 2022, and elicited several questions about the consequences of the suggested design.
But building a trust mechanism for web clients becomes more difficult if people do not trust the entity creating the technology.
WEI was discussed at the W3C Anti-Fraud Community Group in late April and has been published to the web as part of the normal iterative process through which browser capabilities get developed.
Despite the spec's half-baked state, the blowback last week was swift – in the form of a flood of largely critical comments posted to the WEI GitHub repository, and abuse directed at the authors of the proposal. The Google devs' response was to limit comment posting to those who had previously contributed to the repo and to post a Code of Conduct document as a reminder to be civil.
The concerns raised include: potential violation of EU data rules; all web interaction would be subject to attestation – something Google explicitly rejects; barriers to new browsers; general distrust of Google; worries about DRM for the web; possible limitations on ad blocking; and more.
- Google asks websites to kindly not break its shiny new targeted-advertising API
- Google snubbed JPEG XL so of course Apple now supports it in Safari
- Google searchers from years past can get paid for pilfered privacy
- Google IO: A deeper dive into the developer day's details
Jay Freeman (aka "Saurik"), the developer of Cydia for jailbroken iOS devices, described the proposal in an online post as the "inevitable end-game of the web" under ad-based business models.
In an email to The Register, Freeman said assumptions about the web being an open standard under which anyone can build a compliant browser have been breaking down for a while now because the software has become so complex.
Because more and more functionality keeps being added, he said – which web publishers in turn expect – there are only a few browser implementations that have kept up.
"If websites are going to require 'this is proven to be one of a small, trusted set of browsers – unmodified from their original behavior – that we believe will, in fact, show our ads to a real user,' then the bar only goes up for building a new web browser."
But wait, there's more
Freeman contends WEI is more than just another barrier to building a competitive browser.
"I feel like there is something even bigger at stake: this takes away even more control over your computer," he argued. "The only reason this is even possible is due to DRM technology sitting on most people's computers, such as Arm TrustZone and Intel SGX.
"Elon Musk right now wants everyone to use only official Twitter apps to talk to his service, and Reddit recently went in a similar direction: exposing trusted computing primitives to apps means that they could ensure that only official clients access their sites. If Google does push this agenda, I thereby believe this would be one of the biggest attacks on not just the open web but on the basic freedom to run a general purpose computer we have so far seen: you can't trust the browser on an 'untrusted' OS."
This would be one of the biggest attacks on not just the open web but on the basic freedom to run a general purpose computer we have so far seen
Freeman added, "I do believe Google is at least being honest in their use cases … they are just slanting them in ways that make me upset: publishers have a want for their ad-based business models to work, and they thereby would like to have a way to require users to only use trusted browsers that will comply … while this spec makes it sound like users are demanding the ability to prove to publishers that they are in fact not running an ad blocker."
In a post on Monday, Brian Grinstead, senior principal engineer for web platform at Mozilla, expressed opposition to the proposal.
"Mozilla opposes this proposal because it contradicts our principles and vision for the Web," he wrote. "… Detecting fraud and invalid traffic is a challenging problem that we're interested in helping address. However this proposal does not explain how it will make practical progress on the listed use cases, and there are clear downsides to adopting it."
Among those familiar with the way browser technology gets developed, Alex Russell, partner product manager on Microsoft Edge and former senior staff engineer at Google, took to Mastodon to urge people to withhold their judgment until WEI is more fully developed.
"Particularly in the early design phase, lots of ideas are bad!" Russell said. "And that's OK! API design requires a journey through a problem space, and the best way to redirect this sort of thing isn't to extrapolate to worst-case scenarios, it's to ask that folks show their work and demonstrate value."
Chris Palmer, a former Google engineer who now works at Tailscale, last week called the proposal a bad idea in a separate Mastodon post.
"Remote attestation misaligns incentives wildly," he wrote. "If you make your customer your enemy, you have profoundly screwed the pooch. A framework for enabling publishers to make their customers their enemies is a framework for profoundly screwing the pooch."
There's no tweaking to fix it. Just drop it and apologize
Ondřej Pokorný, a freelance technology consultant, offered similar sentiment, via Mastodon. "The problem with many of these new APIs from the whole 'Privacy Sandbox' and other proposals intended to replace 'legitimate' third-party use-cases is that it's turning the browser from a User-Agent into double agent working also in the interest of advertisers and other corporate players, often not aligned with user interests," he argued.
Palmer added, "The best outcome is for Google to simply retract this proposal tomorrow morning. There's no tweaking to fix it. Just drop it and apologize."
The Register asked Google to comment and the web goliath declined. However, we understand it intends to address concerns and supposed misapprehensions raised about the proposal in a forthcoming message. ®