The safe_eval function in Ansible before 1.5.4 does not properly restrict the code subset, which allows remote attackers to execute arbitrary code via crafted instructions.
"https://github.com/pypa/advisory-database/blob/main/vulns/ansible/PYSEC-2020-199.yaml"