Apple MobileMe Brief

July 2011


OpenNet Initiative researchers examined allegations that email messages sent from Apple’s MobileMe service are being censored because of sensitive political content. This brief concludes that Apple is not censoring because of political content; messages are likely being dropped because of overzealous spam filtering. However, this issue raises questions about the impact of aggressive spam filtering, the lack of notification to users and the accountability of service providers.


On June 30, 2011, a user on Apple support group forum MacInTouch reported1 that outgoing emails sent through Apple’s MobileMe2 service were being filtered without warning. The user experimented with a variety of different messages and found that after removing the phrase “growing hostility against Frankfurt and Brussels”, the message was delivered normally.

The next day a fellow member of the forum attempted to send a email describing politically sensitive events (“Syria, Greece, Saudia Arabia and Yemen continue to oppress and massacre their own citizens”) through the MobileMe service. Again, this user claimed that the message did not send successfully. Other users of this forum expressed concern that Apple appeared to be censoring email messages based on political content.

This story was first reported widely in a July 6th article in The Register, which described other users who had duplicated these results.3 Several other reports4 claimed to have also duplicated these results, although in some cases the failure to deliver messages appeared to be intermittent.5

Apple responded to these claims on July 7th, 2011. An Apple spokesperson told Talking Points Memo that “Apple is not blocking e-mails with political content. Occasionally, automatic spam filters may incorrectly block legitimate e-mail. If a customer feels that a legitimate e-mail is blocked, we encourage customers to report it to MobileMe support."6 We contacted Apple for comment but did not receive a response.


ONI researchers sent out several messages from a MobileMe account to a Gmail account, with a carbon copy sent to the same MobileMe account. Testing was conducted on both the web-based client at and a Thunderbird email client configured to send mail via

Figure 1: A message that did not send successfully from a MobileMe account

The message was then systematically adjusted, using different country names, changing negative words to positive variations, and altering the grammatical structure of the messages. Nearly all versions continued to be blocked. However, we found that a version of the message which made less sense grammatically but contained all the relevant keywords did get through, using both the web interface and Thunderbird. (See Figure 2)

Figure 2: A message successfully sent from a MobileMe account

The two messages seen in Figures 1 and 2 were then sent with a second MobileMe account, and the exact same results were observed. As an additional test, the original, blocked email seen in Figure 1 was sent from a Gmail account to the MobileMe account. The message was delivered but flagged by MobileMe as spam. (See Figure 3)

Figure 3: A message sent to a MobileMe account, identified as spam

After clicking the “Not Junk” button and resending the message from Gmail to the MobileMe account, the re-sent message was delivered normally. Attempting to send this original message from MobileMe again still failed to deliver successfully.

The original message seen in Figure 1 was also sent through four other webmail providers:,, and (from a Hotmail account). The message was delivered successfully each time, however in two cases ( and the sender needed to solve a captcha. It should be noted that both of those email addresses are used very infrequently, whereas the address is actively used.

In a further attempt to send the blocked message seen in Figure 1, the text of that message was placed between two paragraphs of text from a recent Information Warfare Monitor report. This message also failed to be delivered, and a subsequent attempt to send only the two paragraphs from the report was also not received successfully. However, we were unable to consistently reproduce this behaviour using other paragraphs from the same report.

As a means of testing non-political content, messages were sent consisting of nothing but words that may be considered objectionable (see Figure 4). This email did not go through.

Figure 4: A message containing nothing but vulgar words. This email was not received.

The same message was then sent with a different subject (“list of naughty words”), and it did go through (see Figure 5).

Figure 5: The vulgar message with a different subject line. This email was received.

Results and Analysis

Two nearly identical messages were sent (See Figures 1 and 2), differing only in two ways: “Their” was changed to “They are” and the word “attempts” was omitted from the second message. The first message was not delivered to recipients while the second one was. In all cases, mail sent from the MobileMe web interface did appear in the “Sent” folder. The following are some possible explanations:

Censoring of sensitive keywords or political content
Very unlikely. The relevant keywords, aside from the word “attempt”, appear in both messages. Other words in the message seem much more likely to trigger any kind of keyword filter. Similar messages with sensitive words and country names altered also failed to send successfully. Other messages consisting of objectionable but non-political words also failed to be delivered.

Similar text previously identified as spam by Apple
It is possible that messages with similar content or structure were previously flagged as spam, triggering the block. The original email that our message was based on had some additional content at the end which could arguably be seen as spam-like:

Stop the oppression of innocent Arab People!!!
Take a stand against Authoritarian Oppressive Regimes!
Equal Rights to all people!

The fact that text identical to the first part of this original message was used may be the reason it was blocked. The original message seen in Figure 1 was reported to be blocked, and it is possible that it resembled a known piece of spam email. We have limited evidence that otherwise innocuous text may be classified as junk if it has been seen alongside known spam.

Delays in processing email
It is possible that these messages have simply been delayed or are experiencing intermittent errors. Apple’s support forums suggest that some messages incorrectly identified as spam may be delayed by up to 72 hours.7 We have not encountered any reports of messages being delivered after this time period, and in our testing this did not occur. We also attempted to send messages multiple times over different intervals to account for intermittent technical issues, so it is not believed that such issues are responsible. Numerous other emails sent between blocked messages did make it to the recipient immediately.

Over-zealous spam filtering
In a knowledge base article, Apple states that they do in fact filter outgoing messages:

If your recipient still does not receive your mail but your mail is being sent without any alerts or "Undelivered mail" messages, then your mail might be blocked or filtered by the mail systems at .MobileMe or the recipient's location.8

Outbound spam filtering by MobileMe is the most likely explanation for the observed behaviour, especially given that the message was flagged as junk when it was received.

This type of overzealous spam filtering is highly problematic when legitimate messages are blocked without the sender being notified. However, it is worth noting that if Apple were to provide such feedback to the user it could help spammers more easily tailor their messages to bypass the filter.

Outbound spam filtering by MobileMe is the most likely explanation for the observed behaviour, especially given that the message was flagged as junk when it was received.

This type of overzealous spam filtering is highly problematic when legitimate messages are blocked without the sender being notified. However, it is worth noting that if Apple were to provide such feedback to the user it could help spammers more easily tailor their messages to bypass the filter.

Apple’s Terms of Service

Our examination of Apple’s MobileMe Terms of Service (TOS) found no explicit mention by Apple of the filtering of spam emails or content. However, under “Section 5. Content and Your Conduct” and “Section 7. Content Submitted or Made available by You on the Service,” Apple’s TOS opaquely states that:

Removal of Content
You acknowledge that Apple is not responsible or liable in any way for any Content provided by others and has no duty to pre-screen such Content. However, Apple reserves the right at all times to determine whether Content is appropriate and in compliance with these TOS, and may pre-screen, move, refuse, modify and/or remove Content at any time, without prior notice and in its sole discretion, if such Content is found to be in violation of these TOS or is otherwise objectionable.

Changes to Content
You understand that in order to provide the Service and make your Content available thereon, Apple may transmit your Content across various public networks, in various media, and modify or change your Content to comply with technical requirements of connecting networks or devices. You agree that the license herein permits Apple to take any such actions.

Apple’s TOS explicitly states that it reserves the right at all times to not only pre-screen MobileMe services to determine the appropriateness of content, but also remove and modify content it deems objectionable at any time. Moreover, the TOS mentions that content may be transmitted over public networks and modified or changed by Apple in order to comply with technical requirements. Apple neither specifies what these technical requirements are nor the public networks that content is transmitted over.

In addition, it states that MobileMe is not to be used to “upload, download, post, email, transmit, store or otherwise make available any Content that is unlawful, harassing, threatening, harmful, tortuous, defamatory, libelous,abusive, violent, obscene, vulgar, invasive of another’s privacy, hateful, racially or ethnically offensive, or otherwise objectionable.”

While these terms are fairly standard as companies’ terms of services, MobileMe’s TOS fails to clearly outline or provide any indication as to how objectionable content is evaluated and determined by Apple. Combined with the lack of notification, the ensuing result leaves MobileMe users’ guessing as to whether an email they have sent a friend or family member will be received or classified as spam. If a MobileMe user were to send a message similar to the text seen in Figure 1, it would not be received despite the email not violating Apple’s TOS.

Is Apple violating their own TOS here? Our findings indicate that they are. In examining MobileMe’s TOS, we are left with more questions than answers on the issue of email filtering. For example, how does Apple determine the appropriateness of content and what is or is not objectionable? How is content modified or changed in order to comply with technical requirements? Is a user’s privacy compromised by Apple’s decision to transmit content across public networks and the right to pre-screen content?

Suggestions for Further Research

There are a variety of possible avenues for future research in this area. We highlight three:

  1. A more systematic comparison of how different service providers deal with spam could include these providers’ terms of service, notification practices and possibilities for feedback and accountability.
  2. A comparative review of providers that focuses on how the variation in spam filtering practices leads to different user experiences.
  3. An interrogation of the various spam filters currently deployed by email providers.


Despite reports of messages sent through the MobileMe service being blocked due to political content, this brief concludes that it is most likely due to overzealous spam filtering. While messages containing certain sensitive political phrases were blocked, other slightly re-phrased versions of the same messages were delivered normally. In addition, we observed messages being blocked that did not contain sensitive political content.

However, the behaviour documented in this brief does demonstrate some problematic aspects of aggressive spam filtering. Broadly speaking, this issue illustrates the difficulty of balancing security and usability of Internet services with the need for a free and open Internet. More specifically, it raises questions about the challenges users of Internet services can face when terms of service are strictly enforced, potentially without notification or possibilities for recourse. In the case of MobileMe, overzealous spam filtering led to legitimate email messages being blocked without notification, and with few options available to users to resolve the problem. Even if such blocking is not due to political censorship, the net effect to users is a diminished ability to communicate without notification or the possibility for recourse.

Appendix A report first describing issue: [Only accessible through google cache]

June 30, 2011
Mike Conley

I haven't seen much in the way of coverage of this particular problem, so I thought I would share my experience, which will lead to my cancelling my eWorld/.Mac/MobileMe membership in the near future after over 10 years -- maybe 15, I can't remember.

It's well-known that MobileMe mail last year removed the ability to disable the Junk filter on incoming mail. This is a nuisance for those of us who don't use Mail (or IMAP), and who have our own spam filters which work quite well, thank you very much. It means that at least once every day or two I have to log onto my MobileMe mail account directly to check for any valid messages that have dropped into the Junk folder -- and there often are such messages.

That was bad enough. Recently, however, I've discovered that MobileMe is filtering *outgoing* E-mail.
When I send E-mail through my client app (MailSmith), I automatically BCC myself. That way, I have a copy of all my sent mail -- and I know whether it actually got sent, since the BCC comes from the SMTP server, and is not just a copy made locally by the app.

On at least 3 occasions (that I know of), an E-mail message I've sent has not been directed to its recipient(s) by the MobileMe SMTP server. This is not a simple glitch; I've resent the exact same messages several times, and they never make it through in their original form. If I edit the messages and remove certain words or phrases, they do get sent. It doesn't matter whether I use Mailsmith, Mail, or the MobileMe Web interface: messages containing the original text do not make it past the server.
In one case, removal of the phrase 'growing hostility against Frankfurt and Brussels' allowed an E-mail message which had been blocked to be sent. Most recently, an E-mail I was sending regarding the growing unrest in Greece has been blocked.

It's fairly clear to me that Apple is silently filtering outgoing E-mail; no notice is given when a message is dropped. On what grounds they feel it reasonable to do this and using what criteria are not clear to me, but, as far as I'm concerned, there is simply no justification for this policy.

I contacted Apple support about this problem, and the tech on the other end went away for a day and eventually came back and told me the message I wanted to send would now make it through. And it did. But they did not do what I requested, which was to disable the outgoing filtering, because a couple of weeks later, the same thing has happened.

I intend to contact Apple support again, but this will be the last time. I intend to find an independent server and obtain my own domain and services independent of Apple, for the first time in over a decade. This is unfortunate, as I've rather enjoyed having my .mac E-mail address, but given Apple's recent authoritarian tendencies in this and other areas, I no longer see any reason to continue supporting them except where absolutely necessary.

    Filtering Types: