The `open` command on macOS doesn’t open websites if no schema is specified. At best (worst?) it makes a suggestion:
$ open google.com
The file /private/tmp/google.com does not exist.
Perhaps you meant 'http://google.com'?
But it doesn’t even try it with the zip TLD: $ open familyphotos.zip
The file /private/tmp/familyphotos.zip does not exist.There's also some that are less commonly encountered by non-tech people, but still fairly common in the tech world (depending on the person of course): .sh, .rs, .cab, .java, .py, .org, .pl, .pm, .target, .pub
That's just from a quick check from what I recognized from memory; I didn't feel like writing the code to also print out the MIME type and description and there may be more. Here's a full list of conflicts I generated from my /usr/share/mime/globs and the TLD list from ICANN, 74 conflicts in total: https://gist.github.com/arp242/ca24b58eaf37184b03e626c8e4093...
Is .zip more common? No doubt, so it does become worse, but this confusion already exists, and will no doubt grow in the future with further proliferation of tlds and file extensions.
Basically, we should probably do something about this regardless, although I'm not entirely sure what.
Of course we all know what will really happen is Google will continue to use Chrome to obfuscate the way the internet works and all of the big tech companies will sell "solutions" to the problems they're causing. Most people will end up completely dependent on Google's Advanced Protection Program, Microsoft's SmartScreen Filter, etc..
Yes, the solution is giving users adequate information. That means telling them whether it's a link or a file, whether the file will execute, or what program will open it.
I think this is a "further research needed". For starters, how often will these situations actually occur in the first place? This may be less often than feared. How many users will be confused? This may be less than one might assume. Is just downloading a .zip file actually a security risk or "merely" confusing? That would make a big difference. What design will help with that? This may be counter-intuitive.
Did you and the GP just invent a new kind of phishing? Lmao. Go grab bitcoin-wallet.zip and start emailing people.
Hi. Here's the cold wallet with the $5 million you requested. The password is 'password'.
Then 5 minutes later send a panicked looking email. Do NOT open the previous email. It was sent to you by accident. You are NOT authorized to view it. DELETE IT NOW!
Haha.Just register those domains and sit back and wait for people to come knocking on your door. It's a phishing dream!
Am I missing something? What does the TLD add that wasn't possible before? Auto-vivified links in the 0.001% of users who only read text/plain? C'mon.
Don't worry. Google will offer enhanced protection to keep your account from being phished and hijacked. How convenient.
Under what circumstances would a link you click on a website or an email _ever_ be a trusted file?
> I've attached familyphotos.zip -- here you go!
The email client will presumably have a link below called familyphotos.zip, but the email client ALSO link-ified the text familyphotos.zip to download the content at https://familyphotos.zip, which is untrusted but you don't realize it doesn't go to the same familyphotos.zip file.
Screenshot: https://twitter.com/mholt6/status/1657133439546695680
Update: Turns out it just contains what_happened.txt, a copy of which is below. Thanks anonymous person!
```
"Hey, this isn't family pictures!"
You're right -- and that link you clicked wasn't a file attached to the email or message you received.
Thanks to Google[0][1], now it's impossible to discern the difference between a link to an attachment called "familyphotos.zip" and a link to this file... unless you are able to inspect the destination of a link before clicking it. Most software and apps don't allow that, and most people don't know how to tell the difference anyway.
Have fun in the Wild Wild Web!
[0]: https://www.blog.google/products/registry/8-new-top-level-do... [1]: https://twitter.com/Google/status/1653866291692728320
```
And most applications (even some browsers) will even helpfully strip out the "ugly https" in front too, and make it clickable. :)
(go ahead, click it)
Depending on platform, this is even more fun with command lines with commands like: `open familyphotos.zip`
But the point is, unless you are technically skilled, you probably can't tell whether you're about to go to a trusted file or download something random.
Depending on the context you would probably expect to be taken to a local file on the disk, you would never think this would download something new unless you were already in a context to do so.
Note that Google is also doing this for .mov. Both are file formats which can execute code (zip and mov both have exploits).
These TLDs should at least be removed from the PSL.