Security Week 18: cryptic hysteria

It was a hard week for the means of exchanging encrypted messages. The means seemed to work, but in one they would find a hole, then in the other there was some other problem. It all started on Monday, when a group of European researchers announced serious vulnerabilities in a number of email clients that support encryption according to the OpenPGP or S / MIME standards ( news ). Announced, shared information with particularly important people "under the embargo": in the world of vulnerabilities they do this so that vendors who fall under the distribution can, for example, release a patch. And also - so that “especially important people” spoke in favor of the importance and necessity of research in public order.

But something went wrong, and instead of Tuesday we had to make public the information on the same Monday (apparently, the discussion went on Twitter and the details began to be revealed). For two published attack scenarios, having a brand (Efail), a logo and a web page looks strange, but, by the way, okay. Both methods of attack are maintained by the message viewer built into mail clients in the form of web pages. A broken img tag is inserted into the message in front of the encrypted text - so that not only the request to download the picture is sent to the attacker's server, but the decrypted message is also attached to it.

In the first attack method, a very simple scheme is used, when a “poisoned” tag is inserted in plain text. In the second, everything is a little trickier: the lack of integrity checking in the same S / MIME (OpenPGP is affected, but to a lesser extent) is exploited, so that a similar “malicious” tag can be inserted directly into the ciphertext.


Video demonstration of attack
')
That is, the attacker must not only gain access to the encrypted messages of the victim. He will have to modify them, send them back to the victim and hope that the built-in "beacon" will work on that side. Not the most effective method, but, on the other hand, we are talking about encryption, and in encryption, the level of paranoia everyone chooses for himself. The EFF Foundation’s not quite adequate response is also interesting: on Monday, having read the report under the embargo, they wrote that “this is a very serious problem,” and recommended that everyone switch to Signal. After the release of the full report, the details were retold on the EFF website , but the recommendation to install Signal (and also remove the plug-ins for decryption) remained.


Experts are joking

As for “remove plugins” and “do not decrypt already received messages”: in the case of Efail there is a temporary solution (until the patch is issued for the mail client you are using) where it is much simpler: turn off the HMTL presentation of messages. In the painttexts, the attack is obviously powerless. And about Signal: the desire of EFF to advertise this messenger on and without a reason is not very clear, especially since the change of infrastructure (mail -> chat) like this, from one small problem, should not happen.


Then there are generally productive negotiations: “We repaired! “Well, no!”

Signal also found a problem, and also not the worst in the world. The researchers found in the desktop version of the messenger excessively arbitrary handling of HTML-codes, which allows for XSS-attack and in general can in every way mislead users. The problem was patched at first, but then it turned out - not completely. And for the Telegram messenger (more precisely, for its desktop version) , the Telegrab Trojan was discovered, stealing the story of the victims' messages and putting it on the server with open access. In difficult times we live!

In one line:
An interesting development of hardware attack Rowhammer on a personal computer. Nethammer develops the ideas of the Throwhammer remote hardware attack and shows how to modify the data on the attacked system without executing a single line of code on it.

Have Cisco found and deleted another tightly wired password .