W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2018

RE: Issue 948 SC 1.3.5 Identify Input Purpose - autocomplete technique VS Privacy/Security

From: Alastair Campbell <acampbell@nomensa.com>
Date: Tue, 5 Jun 2018 21:12:54 +0000
To: John Foliot <john.foliot@deque.com>
CC: WCAG <w3c-wai-gl@w3.org>
Message-ID: <AM5PR0902MB2002DD0A65BE275ABF656420B9660@AM5PR0902MB2002.eurprd09.prod.outlook.com>
Thanks John,

I’m fairly sure the browsers also ask whether to *save* the information in the first place, right?
The commenters point was more about people not realising they were saving information, which would then be available to the next user.

In a library / school scenario with shared computers, basic IT practices mean you should either have your own profile, or it doesn’t save any data between sessions (i.e. things like autocomplete are turned off, and browser-history gets wiped like in a private window).

In an in-home scenario (e.g. domestic abuse) it is your browser history that is the problem, auto-complete does not give away which forms you have or haven’t filled in (correct?).



From: John Foliot

​Additionally, in all of the browsers I've tested (Chrome, Firefox, Vivaldi, Brave) on the Windows platform​, the browser *offers* to auto-complete the fields, but awaits final confirmation from the user - there is no obligation to do so however, and it remains a conscious choice to accept the auto-filling. Additionally, most (all?) browsers allow to set up more than one 'profile', for those 'shared' instances where more than one user's data is stored.

Also, both Internet Explorer and MS Edge do not support autocomplete as currently spec'd, so if a user is concerned about this attribute, they could choose to use a different browser (weak come-back, I know, but true...)

Finally, there is an additional "super" value for autocomplete (off/on) which you would think could be used to "over-ride" specific values (but doesn't, at least not in my quick testing in 3 browsers). The current spec states:

If the autocomplete<https://www.w3.org/TR/html5/sec-forms.html#element-attrdef-autocompleteelements-autocomplete> attribute is omitted, the default value corresponding to the state of the element’s form owner<https://www.w3.org/TR/html5/sec-forms.html#form-owner>’s autocomplete<https://www.w3.org/TR/html5/sec-forms.html#element-attrdef-autocompleteelements-autocomplete> attribute is used instead (either "on<https://www.w3.org/TR/html5/sec-forms.html#attr-valuedef-forms-autocomplete-on>" or "off<https://www.w3.org/TR/html5/sec-forms.html#attr-valuedef-forms-autocomplete-off>"). If there is no form owner<https://www.w3.org/TR/html5/sec-forms.html#form-owner>, then the value "on" is used.

Might be able to file a bug there (as one state not accounted for, where the "form owner" [aka the parent <form> element] is explicitly set to "off", has not been accounted for). I would suggest that given it's a parent element, that the traditional "cascading" would apply (off at the parent level = off at all the child levels as well), but that will need to be discussed at WebPlatforms WG first.


On Tue, Jun 5, 2018 at 4:16 PM, Alastair Campbell <acampbell@nomensa.com<mailto:acampbell@nomensa.com>> wrote:
Hi everyone (and particularly John & Lisa),

I’d like to run a proposed response past the group before posting to github (and notifying the commenter before the group gets a chance to review).


I’d summarise the core issue as: using autocomplete/autofill could be an issue for privacy/security for people using shared devices (e.g. family computer), and autcomplete shouldn’t be proposed as a technique to fulfil it.

You can read the back and forth on the thread, but I’m proposing the response is:

The working group have considered the security and privacy aspects of this, and whilst it must be acknowledged there may be some circumstances in which a user would not want fields identified and auto-filled, the working group feel the benefits outweigh the risks.

Mitigating factors include:

- This is functionality that is already available in user-agents, and used by some websites already.
- It is something that must be enabled within the user-account and browser of the device used.
- People can use various privacy features if that is a requirement.

Currently the autocomplete attribute (for autofill) is the best supported method, so that will be the first technique provided.

Personally, I don’t see it as an issue, but I’d appreciate a review from others familiar with autocomplete.

Kind regards,


John Foliot
Principal Accessibility Strategist
Deque Systems Inc.

Advancing the mission of digital accessibility and inclusion
Received on Tuesday, 5 June 2018 21:13:26 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 21:08:26 UTC