The images in this article were generated with artificial intelligence. How we publish
Cisco warned about a critical vulnerability in AsyncOS for its Secure Email Gateway (SEG) that is already being exploited in real environments. The failure, identified as CVE-2026-76461 and with a CVSS score of 9.8 / 10, allows a remote and unauthenticated attacker to send a specially manipulated email whose processing can lead to the execution of arbitrary commands with root privileges in the underlying operating system.
According to the official notification, the root of the problem is an insufficient validation in the logic that processes and parses incoming messages. Cisco describes that a malicious message may contain exploitable SQL sentences; the execution of those sentences could result in system-level execution of commands. Done confirmed: Cisco has published corrected versions and states that the active exploitation has been detected in nature; in addition, the United States Cybersecurity and Infrastructure Agency (CISA) added the CVE-2026-76461 to its catalogue Known Exploited Vulnerabilities (KEV), imposing the obligation to apply patches in federal agencies before 17 September 2026. The CVE tab and the CISA entry are available at the following links: NVD - CVE-2026-76461 and CISA KEV - CVE-2026-76461.

Technically, Cisco describes the vector as the insertion of SQL sentences into the mail flow that are not correctly neutralized prior to execution by internal components of the SGE. As a useful indicator given by Cisco, the mail service text records (mail _ logs) must be reviewed for patterns that match COPY. * TO PROGRAMME, since the presence of that chain in the logs may indicate attempts to use SQL instructions that invoke the operating system. The company provided the recommended search for inspection:grep -i "COPY. * TO PROGRAMME" [route _ to _ mail _ logs](on Cisco teams that default route is mail _ logs).
Products concerned (confirmed): Cisco Secure Email Gateway, in both physical and virtual versions, regardless of the device's configuration. Cisco explicitly indicated that other products such as Secure Email and Web Manager and Secure Web Application are not affected.
Available solutions (confirmed): Cisco published patches / fixes in the following branches of AsyncOS: for the 15.5 series (corrected in 15.5.5-0141), 16.0 (corrected in 16.0.4-302) and 16.5 (corrected in 16.5.0-780). The official note notes that there are no complete alternative solutions: the only effective measure is to update to a corrected version.
Cisco claimed to have directly contacted customers whose cloud devices showed malicious activity, but did not disclose the numerical scope of the incidents. Done confirmed: Cisco warns that, if the operation is successful, the attackers will obtain command execution with root privileges and could remove or alter evidence within the affected device. For this reason, the company also recommends reviewing external records (network and firewall looms) that cannot be manipulated from the compromised SEG to discover abnormal activity, uploaded from the device to external IP or downloads from malicious directions.
There are aspects that remain publicly uncertain at this time: Cisco has not published detailed technical details of the explosion used, nor has it disclosed complete network commitment indicators (IoC) or the total number of devices affected. Nor has the family or identity of the actors who are exploiting the failure been confirmed.
From a practical technical perspective, the reference to the pattern "COPY... TO PROGRAMME" suggests a vector in which SQL sentences with the ability to invoke operating system processes are used as a lever to scale to command execution. Reasonable estimate (not confirmed by Cisco): In some SQL engines, the COPY TO PROGRAMME instruction allows you to direct database data to the execution of a system program; if the SEG component processes SQL inputs directly without sanitation, that behavior could be the path to running shell commands. This explanation is a technical interpretation of the indicators that Cisco did confirm, not an official statement on internal architecture.
What managers and security officials should do (concrete and prioritized steps):
1) Instant. Update any Cisco Secure Email Gateway affected to one of the corrected versions indicated by Cisco as a priority. If possible, make the update out of production hours with a rollback plan.
2) Inspection of local login. Run the search provided by Cisco on all mail _ logs and on each cluster device: grep -i "COPY. * TO PROGRAMME" mail _ logs (adjust the route according to your installation). Any coincidence should be considered a possible compromise.
3) Review external records and network indicators. Because a root actor can erase traces, cross the SEG records with firewalls, proxys and network sensors to identify unexpected outgoing connections, file transfers or remote accesses initiated from the SEG.
(4) Containment and recovery if there is evidence of exploitation. If you find evidence of commitment, isolate the network device, preserve a forensic copy of the external log and plan a complete reimage from reliable installation means or restoration from known clean backups. Do not trust to delete only processes or files: root climbing usually requires reconstruction.
(5) Rotation of credentials and review of integrity. Change keys and credentials associated with the affected device (API keys, administrative passwords, certificates) and check the integrity of the adjacent accounts and systems that may have been accessed after an intrusion.

(6) Monitoring and continuous detection. Add rules to SIEM / IDS to detect operating-related patterns (search for suspicious SQL chains, traffic to unusual PIs) and coordinate with your Cisco supplier / tac to confirm findings and receive additional mitigation.
For US public sector organizations and contractors, remember that inclusion in the KEV catalogue implies a regulatory obligation to apply mitigation within the time limits indicated by CISA. In addition, if you need public references about the failure, see the entry of the CVE and the CISA page mentioned above; for patch details and final mediation steps, use Cisco's official communications and technical support.
Conclusion: This is a high-gravity and active operation failure that affects devices that process incoming mail and which, by its nature, can lead to a complete system take. The only reliable measure known is to apply the corrected versions; in the meantime, do an immediate search on the logs, monitor outgoing communications and prepare for forensic and reimage procedures if evidence of intrusion appears.
Related
More news on the same subject.

GhostAction Expands Attack: phishing to steal credentials and persistence in VS Code
A group behind the GhostAction campaign - already known for infiltrating GitHub Actions workflows into open source projects - has expanded its arsenal: in addition to installing...

OpenCode: possible exposure of AI suppliers in Console affects 1.4% of users
On the night of October 16 there was an unauthorized access to a system inherited from OpenCode which, according to the company, could have exposed keys from IA suppliers that u...

ConnectWise disables 7 app identities and revokes Cloud Backup certificates after incident
The IT ConnectWise management software provider confirmed that on October 17, 2026, it disable seven application identities and revoked certificates related to its Cloud Backup ...

Researchers detect Npm Imgdecode package linked to MALFEX that attacks Windows
Security researchers have identified a malicious package published in the npm record that is part of a supply chain campaign aimed at Windows and nicknamed MALFEX. The package, ...

OpenBSD 7.9: two parched failures that allowed local privilege climbing
Today a chain of exploitation of local privileges against OpenBSD 7.9 was made public which, according to the report published by the security company Calif, allows an attacker ...

OpenAI reveals unaligned IA incidents; a GPT-5.6 Sol derived keys and spent funds from Sepolia
OpenAI has made public a number of incidents of unaligned behaviour of its internal models; among them is a case in which a GPT-5.6 family research model Sol derived private key...