Neither Snow Nor Rain Nor MITM...

Abstract

The SMTP protocol is responsible for carrying some of users' most intimate communication, but like other Internet protocols, authentication and confidentiality were added only as an afterthought. In this work, we present the first report on global adoption rates of SMTP security extensions, including: STARTTLS, SPF, DKIM, and DMARC. We present data from two perspectives: SMTP server configurations for the Alexa Top Million domains, and over a year of SMTP connections to and from Gmail. We find that the top mail providers (e.g., Gmail, Yahoo, and Outlook) all proactively encrypt and authenticate messages. However, these best practices have yet to reach widespread adoption in a long tail of over 700,000 SMTP servers, of which only 35% successfully configure encryption, and 1.1% specify a DMARC authentication policy. This security patchwork -paired with SMTP policies that favor failing open to allow gradual deployment -exposes users to attackers who downgrade TLS connections in favor of cleartext and who falsify MX records to reroute messages. We present evidence of such attacks in the wild, highlighting seven countries where more than 20% of inbound Gmail messages arrive in cleartext due to network attackers.

What Email Servers Can Tell to Johnny: An Empirical Study of Provider-to-Provider Email Security

“…As observed from table 5, for SPF, the various results are quite close except the one pertaining to the Alexa dataset in [51], which however seems more natural for the time that research was conducted. Regarding DKIM, the high percentage reported by [51] is probably due to the skew toward major email providers as noted by the authors in the Gmail dataset. Lastly, with reference to our findings, the DMARC score has almost doubled if compared against that in [51].…”

“…Regarding DKIM, the high percentage reported by [51] is probably due to the skew toward major email providers as noted by the authors in the Gmail dataset. Lastly, with reference to our findings, the DMARC score has almost doubled if compared against that in [51]. This increment seems reasonable if we consider the time span between these contributions.…”

“…Given that RFC 8446 [50] defines the AES_128 or AES_256 in GCM or CCM mode with SHA_256 or SHA_384, or the CHACHA20_POLY1305_SHA256 ciphersuites for use with TLS 1.3, it is concluded that ≈89.5% of the inbound MTAs in the dataset did apply a recommended ciphersuite, while an additional ≈7% a secure one. Indicatively, the work in [51], which is further considered in section V, reported a 51.5% usage of the TLSv1.2 ECDHE_RSA_AES_128_GCM_SHA256 ciphersuite for Gmail inbound traffic on Apr. 30, 2015.…”

“…An ≈2.5% employed weak keys having a length less than 1024-bits. The work in [51], reported a 86.4% use of 2048-bit keys based on April 2015 data. Naturally, the improvement of ≈9% for keys equal or greater than 2048-bits is on the positive side, but not remarkable considering the five years distance.…”

“…About one year latter, the authors in [51] presented an empirical analysis of STARTTLS, SPF, DMARC and DKIM based on two datasets, namely Google transparency report (Gmail logs from Jan. 2014 to April 2015) and the April 2015 Alexa list of top 1M ranked websites. Their work included a study on DNS Hijacking, but neither DANE nor DNSSEC usage by SMTP servers were considered.…”