MIT Kerberos 5 (krb5) contains a flaw related to the acceptability of checksums. This may allow a remote attacker to modify user-visible prompt text, modify a reponse to a KDC, or forge a KRB-SAFE message via unkeyed checksums or the use of RC4 keys.
Currently, there are no known workarounds or upgrades to correct this issue. However, MIT has released a patch to address this vulnerability. Check the vendor advisory or solution in the references section.