I understand we have all the proof now, but back then, how can someone get intimidated by a call that could easily be fake.
I understand we have all the proof now, but back then, how can someone get intimidated by a call that could easily be fake.
Fair enough, but a decent actor could probably work as well.
How did he prove he was actually himself? Seems like anyone could have done it with a bit of AI.
For that they use iframes, which have a different security system.
Because of the CORS settings on Google’s servers would tell your browser to not go forward with the request. There are two ways it could eventually be possible:
Fair enough, that’s interesting. I assume this only applies to the non-web clients. On the web, it would not be possible. You can verify by looking at the outgoing network requests on this random video for example: https://invidious.privacyredirect.com/watch?v=qKMcKQCQxxI
I’m pretty confident that you are wrong.
Invidious and YouTube piped (and LibreTube) by default load the videos server-side, as opposed to GrayJay, NewPipe or Smarttube.
It has advantages (mostly that your IP address is not shared with YouTube, and it allows users from countries where YouTube is blocked to still access it) and inconvenients (much harder to keep up when YouTube actively seeks to block them).
You can download videos and cut off sponsored moments in the video with sponsorblock.
GrayJay is pretty good!
LibreTube is also a good one. Basically an app for piped
Browsers based on chromium do not have to follow exactly what the main branch is doing. If they want to keep supporting MV2 or support different rules for MV3, they can. Albeit it’s a bit cumbersome.
Unfortunately, I think that while ad blockers won’t work as well, they will still work good enough that most won’t bother making the switch.
https://blog.getadblock.com/how-adblock-is-getting-ready-for-manifest-v3-6cf21a7884f6
https://github.com/uBlockOrigin/uBOL-home/wiki/
https://adguard.com/en/blog/adguard-mv3.html
https://www.reddit.com/r/uBlockOrigin/comments/1067als/comment/j3h00xj/
The main issue I see is the slow update of filters (which require an extension update). This might make YouTube win the cat and mouse game. Where YouTube updates(ed?) their blocking detection multiple time a day.
Firefox’s implementation of manifest v3 doesn’t come with the same restriction as Google’s. Ad blockers will still work with manifest v3 on Firefox (but not on chrome).
This means that all manifest v3 extensions made for Chrome work with Firefox, and almost all manifest v3 extensions made for Firefox will work with Chrome.
They already support manifest v3, but with less restrictions than Chrome’s implementation.
Firefox’s implementation of manifest v3 is a bit different than Chrome’s, and still allows for blocking webrequests with no upper limit.
https://blog.mozilla.org/addons/2024/03/13/manifest-v3-manifest-v2-march-2024-update/
https://blog.mozilla.org/addons/2022/05/18/manifest-v3-in-firefox-recap-next-steps/
It would be crazy expensive to run an attack of this size for years.
There is no way a DDoS on the website in affecting the crawler. Also, running a DDoS attack of this size costs a lot of money (if you rent the network, if you own it it costs money as lost sales). No one is giving AI control over a DDoS network to just fuck around.
It works on Android, but I don’t believe it works on iOS.