Cisco issued an advisory on July 17 concerning severe vulnerabilities within its Identity Services Engine (ISE) and ISE Passive Identity Connector (ISE-PIC), which could permit an unauthenticated remote attacker to execute commands with root privileges.
Cisco released multiple patches to address these issues, including an expanded fix for specific software versions. The vulnerabilities were initially reported by Bobby Gould of Trend Micro Zero Day Initiative and Kentaro Kawane of GMO Cybersecurity by Ierae, collaborating with Trend Micro Zero Day Initiative. These vulnerabilities enable arbitrary code execution on affected systems.
Three specific vulnerabilities are addressed by Cisco’s patches: CVE-2025-20281, CVE-2025-20337, and CVE-2025-20282. Although all three allow for arbitrary code execution, they are distinct and do not require combined exploitation for effectiveness. CVE-2025-20281 and CVE-2025-20337 affect both Cisco ISE and Cisco ISE-PIC. An attacker could exploit these by submitting a crafted API request, leveraging insufficient validation of user-supplied input, which could result in root-level privileges.
A critical flaw in Nvidia’s toolkit allows AI container escapes
CVE-2025-20282 specifically impacts Cisco ISE and ISE-PIC Release 3.4. This vulnerability could be exploited by an attacker uploading a crafted file to the device. Due to a lack of file validation, the malicious file could be placed in privileged directories, subsequently allowing the attacker to execute arbitrary code or gain root access. Cisco has stated it is not aware of any active exploitation of these vulnerabilities at this time.
Cisco ISE installations are considered patched against these vulnerabilities if they are running Release 3.4 Patch 2 or Release 3.3 Patch 6 (with Release 3.3 Patch 7). While Cisco had previously released hot patches, these have been superseded by the versions listed. The company has also provided documentation detailing the process for applying these updates.
In related cybersecurity developments, Cisco’s security intelligence division, Talos, recently identified a threat actor group utilizing generative AI as a lure to distribute malware. This group employed a spoofed version of a legitimate business’s website to disseminate the CyberLock ransomware, which encrypted specific documents on victims’ computers. The deceptive website offered a downloadable version of ChatGPT as bait. Separately, in March, Cisco initiated a digital skills training program across the European Union through its Networking Academy, offering free courses to enhance individuals’ networking and cybersecurity skills.
