On Mon 02 Nov 2015 at 23:02:38 +0100, Vincent Lefevre wrote: > On 2015-11-02 15:00:19 +0000, Brian wrote: > > On Mon 02 Nov 2015 at 14:58:24 +0100, Vincent Lefevre wrote: > > > > > On 2015-11-02 13:47:41 +0000, Brian wrote: > > > > On Mon 02 Nov 2015 at 14:17:39 +0100, Vincent Lefevre wrote: > > > > > The user's browser cannot compromise the site itself. But a security > > > > > bug may permit an attacker to get the user's login and password, and > > > > > neither the bank nor the user would like this. > > > > > > > > Would this obtaining of the password be before or after encryption > > > > takes place? > > > > > > With an XSS[*] vulnerability, before. > > > > > > [*] https://en.wikipedia.org/wiki/Cross-site_scripting > > > > Quoting from that page: > > > > XSS enables attackers to inject client-side script into web pages > > viewed by other users. > > > > The bank's site would be compromised. It wouldn't matter what user-agent > > string was sent by the user. > > No, the injection happens locally (after the web page is fetched), > in the user's browser, not remotely.
An attacker must inject a payload into a web page that the user visits. When the page loads in the user’s browser the attacker’s payload will be executed. A user would likely have no knowledge of this, irrespective of whatever browser or user-agent string is being used. Without the payload (which the bank's site has delivered) the security of the browser is not compromised. If a password were to be obtained the bank is complicit in the action. I expect they would take responsibilty for this.