Secure shell (SSH) transport layer protocol implementations from different vendors contain multiple vulnerabilities in code that handles key exchange and initialization. Both SSH servers and clients are affected. A remote attacker could execute arbitrary code with the privileges of the SSH process or cause a denial of service.
From the IETF draft SSH Transport Layer Protocol:
SSH is a protocol for secure remote login and other secure network services over an insecure network.
The impact will vary for different vulnerabilities, but in some cases remote attackers could execute arbitrary code with the privileges of the SSH process. Both SSH servers and clients are affected. On Windows systems, SSH servers commonly run with SYSTEM privileges. SSH daemons on UNIX systems typically run with root privileges. In the case of SSH clients, any attacker-supplied code would run with the privileges of the user who started the client program. Additional privileges may be afforded to an attacker when the SSH client is configured to run with an effective user ID (setuid/setgid) of root. Attackers could also crash a vulnerable SSH process, causing a denial of service.
The CERT/CC thanks Rapid7 for researching and reporting these vulnerabilities.