New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 671007 link

Starred by 1 user

Issue metadata

Status: Duplicate
Owner:
Closed: Dec 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux , Windows , Chrome , Mac
Pri: ----
Type: Bug-Security



Sign in to add a comment

Security: Clickjack (or keyjack) users into unknowingly installing extensions

Project Member Reported by rob@robwu.nl, Dec 4 2016

Issue description

Chrome version: 55.0.2883.75 (stable), 57.0.2942.0 (latest)

Web pages can prompt users to install extensions from the Chrome Web Store. This prompt appears in a predictable location in the page and features two buttons: "Cancel" and "Add extension". This UI can be abused by mouse or keyboard:
- Mouse: Get the user to repeatedly click at the expected location of the "Add extension" button, and trigger the installation dialog.
- Keyboard: Trigger the install dialog before the user hits arrow key - spacebar.
(these circumstances are quite realistic in a game)

This is equivalent (or worse) than UXSS because extensions with arbitrary permissions can be installed in this way, without the user's explicit consent.

To cover their tracks, an attacker can clone a popular extension and uninstall the original extension (via the chrome.management API). The average user won't see the difference...


Here is a quick PoC using the real AdBlock Plus (an attacker would obviously use their own site and a malicious extension):

1. Visit https://adblockplus.org/
2. Open Chrome's developer tools, go to the console and run the code from poc-game.js
3. Play the game (e.g. press arrow right, spacebar, arrow right, spacebar).

The attached video shows the PoC in action. Note that the bubble at the end of the video would immediately be hidden if this vulnerability was abused via mouse, because the bubble disappears when the user clicks anywhere.


Suggested mitigation: Add a slight delay (e.g. a second or two) before the user can click on the "Add extension" button.
 
extension-install-trick.ogv
621 KB View Download
poc-game.js
545 bytes View Download
Labels: -OS-All OS-Chrome OS-Linux OS-Mac OS-Windows
Status: Untriaged (was: Unconfirmed)
Summary: Security: Clickjack (or keyjack) users into unknowingly installing extensions (was: Security: too easy trick users into unknowingly installing extensions)
Essentially the same as Issue 636974, except this time the trusted UI is the extension installation UI.

From a security POV, Chrome Web Store reserves the right to disable inline install of extensions (so users must be on the Chrome Web Store website) but that's a reactive measure.
Labels: Security_Severity-Low Security_Impact-Stable
Owner: rdevlin....@chromium.org
Status: Assigned (was: Untriaged)
Assigning to rdevlin.cronin, but maybe someone from the security UX team should take this?
This is the same issue as bug 394518. The simplest way to prevent this is to delay the accept action to make sure the user sees the dialog.
Mergedinto: 394518
Status: Duplicate (was: Assigned)
Merging.  Agreed that we should probably just add a 500ms delay or so.

Comment 5 by rob@robwu.nl, Dec 6 2016

Please cc me on bug 394518. Was it reported externally? The big number looks alarmingly low.

And question for security team: is clickjacking/keyjacking unconditionally low severity, even if it leads to UXSS?

Comment 6 by awhalley@google.com, Oct 26 2017

Cc: luca.erm...@gmail.com

Comment 7 by awhalley@google.com, Apr 27 2018

Cc: mar...@seznam.cz
Project Member

Comment 8 by sheriffbot@chromium.org, Jun 22 2018

Labels: -Restrict-View-SecurityTeam allpublic
This bug has been closed for more than 14 weeks. Removing security view restrictions.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Comment 9 by mea...@chromium.org, Jun 22 2018

Labels: -Security_Severity-Low Security_Severity-Medium
> And question for security team: is clickjacking/keyjacking unconditionally low severity, even if it leads to UXSS?

No, not necessarily. In fact bug 394518 is medium, not low, so this should have been medium as well (still a duplicate though, I'm afraid).

Sign in to add a comment