Dev News Daily ENDE

SANS honeypots record a wave of exploit attempts for the sshole SSH flaw, peaking at 713 in a day

Johannes Ullrich of the SANS Internet Storm Center reported on 8 November that the centre's SSH honeypots are seeing attempts to exploit the vulnerability known as sshole, tracked as CVE-2026-123158, which emerged the week before.

The honeypots run Cowrie and store only the first 255 characters of each SSH password. Ullrich used that limit as a signal: passwords that long are highly unusual in ordinary brute-forcing, so their appearance marks something new. In his table, such attempts were at zero from 26 to 31 October, then 23 on 1 November, 0, 5, 58, 232, a peak of 713 on 6 November, 246 on 7 November and 51 on 8 November, with the last day still incomplete when he wrote.

He cautions that the first long passwords, on 1 November, were probably unrelated. Buggy brute-force scripts sometimes put commands meant for the shell after login into the password field, for example a chain that downloads a bot binary with wget, busybox or curl and makes it executable. According to Ullrich, the first real attempts against sshole arrived on 5 November and used passwords made entirely of the letter A.

The diary lists the most active early sources by IP address and network. The top three sent 48, 44 and 43 attempts and belong to COLT Technology Services, Datacamp and B2 Net Solutions; several of the others are Korea Telecom addresses, and the list also includes Telecom Armenia, Uzbektelekom, Vodafone Libertel and Scaleway networks. Such addresses usually point to compromised or rented machines rather than to the people behind them.

Ullrich expects that a number of these logins succeeded against the honeypots and says he will follow up with details once he has extracted them.

SANS honeypots record a wave of exploit attempts for the sshole SSH flaw, peaking at 713 in a day
SANS honeypots record a wave of exploit attempts for the sshole SSH flaw, peaking at 713 in a day — Dev News Daily

What it means

Administrators who have not yet applied the fix for CVE-2026-123158 should treat this as active exploitation, not a theoretical risk. Logging unusually long password attempts at the SSH layer is a cheap way to spot the pattern on one's own servers.

Primary source
SANS Internet Storm Center
https://isc.sans.edu/diary/rss/33498