In recent news, Google has put forth a proposal known as the "Web Environment Integrity Explainer", authored by four of its engineers. On the surface, it
Actually, they are controlling your graphics driver. If you’re using a custom driver you’ll fail attestation because you have untrusted code in your kernel and/or browser process. I expect this will also fail if you’re using an old driver with known vulnerabilities that allow you to use your own device in unexpected ways.
Your TPM unit in the motherboard has more privileges than you do. It attests to the integrity of the kernel, graphics driver included, and the kernel attests to the integrity of the browser and any peripherals.
Why can’t it? I’m sure Windows allows non-admin processes to get information about active drivers, secure their own memory, and provide attestation that they are secured, provided TPM and secure boot are enabled.
The technology is useless if you can pass an integrity check just by running as admin. The point is that Google has control over what the process is doing and knows if you’re tampering with it. I guess nothing would stop you from making a device’s that uses the hdcp osd support to draw black boxes over ads you find using accessibility information, but if you’re able to modify the page through extensions or developer tools or memory manipulation, then you’re able to make automated API calls, and preventing that is supposedly the whole point of this system.
The reason for using an external device to overlay data on the video signal is that there is a browser API for tracking occlusion. It’s supposed to be used for things like disabling animations of elements that are not visible, but could be unethically used for things like making you pay extra to listen to videos if you don’t have an extra display to put them on.
I don’t know why you think secure computing doesn’t relate to driver control. Drivers run with special privileges and can modify protected memory. This is why people write root kits, and detecting those root kits is one of the primary motivations behind secure computing.
The major point is not so much whether your browser could block ads - your point regarding the browser ultimately having to render each element is true. The problem is that if the web server gets a request from an unattested browser (such as an old version, or one that has an ad blocker installed), it will refuse to serve any content, not just ads.
Regular people will inevitably get frustrated and we end up in scenarios like “<x browser>is bad, it doesn’t work with <y site>” because of this proposal, and more and more people end up switching until you have to use a compliant (Chromium-based) browser to do anything at all on the internet, and Google’s strangehold on web standards solidifies even further.
deleted by creator
Actually, they are controlling your graphics driver. If you’re using a custom driver you’ll fail attestation because you have untrusted code in your kernel and/or browser process. I expect this will also fail if you’re using an old driver with known vulnerabilities that allow you to use your own device in unexpected ways.
deleted by creator
Your TPM unit in the motherboard has more privileges than you do. It attests to the integrity of the kernel, graphics driver included, and the kernel attests to the integrity of the browser and any peripherals.
deleted by creator
Why can’t it? I’m sure Windows allows non-admin processes to get information about active drivers, secure their own memory, and provide attestation that they are secured, provided TPM and secure boot are enabled.
https://www.microsoft.com/en-us/security/blog/2017/10/23/hardening-the-system-and-maintaining-integrity-with-windows-defender-system-guard/
deleted by creator
You said “Google isn’t controlling your graphics driver”
deleted by creator
The technology is useless if you can pass an integrity check just by running as admin. The point is that Google has control over what the process is doing and knows if you’re tampering with it. I guess nothing would stop you from making a device’s that uses the hdcp osd support to draw black boxes over ads you find using accessibility information, but if you’re able to modify the page through extensions or developer tools or memory manipulation, then you’re able to make automated API calls, and preventing that is supposedly the whole point of this system.
The reason for using an external device to overlay data on the video signal is that there is a browser API for tracking occlusion. It’s supposed to be used for things like disabling animations of elements that are not visible, but could be unethically used for things like making you pay extra to listen to videos if you don’t have an extra display to put them on.
I don’t know why you think secure computing doesn’t relate to driver control. Drivers run with special privileges and can modify protected memory. This is why people write root kits, and detecting those root kits is one of the primary motivations behind secure computing.
The major point is not so much whether your browser could block ads - your point regarding the browser ultimately having to render each element is true. The problem is that if the web server gets a request from an unattested browser (such as an old version, or one that has an ad blocker installed), it will refuse to serve any content, not just ads.
Regular people will inevitably get frustrated and we end up in scenarios like “<x browser>is bad, it doesn’t work with <y site>” because of this proposal, and more and more people end up switching until you have to use a compliant (Chromium-based) browser to do anything at all on the internet, and Google’s strangehold on web standards solidifies even further.
It’s possible but not particularly plausible.
Someone always finds a way.