[OTR-dev] Forward secrecy/deniability for long messages with low overhead

[Ileana]

On Fri, 22 Feb 2013 14:36:25 -0300
Sergio Lerner <sergiolerner at certimix.com> wrote:

> I was thinking again about my proposal and found that it has a
> drawback. Suppose an attacker breaks in a computer that is
> communicating with OTR, he can try to guess a plaintext and use it to
> check if the encryption is correct.

Am I missing something?  If the attacker has compromised the computer,
do they not have direct access to system memory containing the
temporary keys, including the plaintext via keyboard hooks?