eSentire's Threat Response Unit (TRU) compiles the following weekly intelligence overview, providing cybersecurity leaders with expert analysis and insights on the most important events of the past week along with important security tips.
Recent Threat Intelligence Advisories
Noteworthy News
UAT-10147: Integrates Agentic AI Into Post-Compromise Operations
Bottom Line: The Chinese-speaking financially motivated group UAT-10147 has been identified using agentic AI to automate attacks against Internet-exposed web servers around the globe.
On August 20th, 2026, Cisco Talos reported on UAT-10147, a Chinese-speaking cybercrime group using agentic AI to automate large-scale server compromise for SEO fraud and data theft. They then deploy a new cross-platform implant, dubbed SPECTRE, with Linux rootkit and Bring Your Own Vulnerable Driver (BYOVD)-based EDR-killing capabilities. UAT-10147 targeted internet-exposed Windows and Linux web servers globally, affecting organizations in government, education, media, technology, and gaming sectors, with observed victims located in Brazil, Bolivia, China, Canada, and Vietnam. From the group's exposed open directory C2, Talos recovered a target list of roughly 170,000 URLs, which the actor had split into smaller batch files to make large-scale scanning more efficient.
After gaining Remote Code Execution (RCE) through publicly disclosed vulnerabilities, the actor typically deploys an automated script to install BadIIS (SEO fraud malware) or a data-stealing payload, or in some cases manually installs a web shell to establish persistence through additional backdoors. On Windows, the group uses the EfsPotato privilege escalation tool, modifies Windows Defender exclusions to hide malware from antivirus scanning, and establishes persistence via scheduled tasks. On Linux, the group chains multiple known local privilege escalation exploits, including CVE-2021-3156 ("Baron Samedit") and CVE-2022-0847 ("Dirty Pipe"), before deploying implants such as NoodleRAT, SPECTRE, and Meterpreter.
The group installed DeepAudit (an AI-driven source code vulnerability scanner) and PentestGPT on its management infrastructure and used AI tooling built around the ysoserial framework to generate complete, documented exploitation workflows for Java deserialization and ASP.NET ViewState deserialization vulnerabilities. The groups newly identified custom implant, SPECTRE, is a cross-platform (Windows and Linux) backdoor written in C that Talos assesses shows signs of AI-assisted development. The Windows variant loads a known-vulnerable driver, which allows the threat actor to unlink EDR callback hooks, blinding EDR agents. The Linux version of SPECTRE also installs a hidden rootkit that lets attackers hide their processes and the rootkit itself from detection tools. The group deployed a variety of other commodity and open-source backdoors throughout the campaign, including Gh0stCringe, QuasarRAT, Meterpreter, NoodleRAT, web shells, and privilege escalation tools such as GodPotato, JuicyPotato, EfsPotato, and RustPotato.
eSentire Threat Intelligence Analysis:
Talos assesses with moderate-to-high confidence that UAT-10147 is among an emerging class of financially motivated intrusion operators leveraging agentic AI systems to operationalize offensive tradecraft at scale. UAT-10147 is a clear signal that agentic AI has moved from a productivity aid to an operational component of intrusion tradecraft. Rather than using AI simply to write scripts faster, the actor integrated it into iterative exploit refinement, validation, and troubleshooting workflows, effectively allowing a less-skilled operator to execute complex, multi-stage attacks that would traditionally require significant manual expertise.
The SPECTRE implant's BYOVD capability is particularly significant from a defensive standpoint. By loading a legitimately signed but known-vulnerable driver, the malware gains kernel-level read/write access it uses to unlink EDR callback hooks entirely, rather than simply terminating the security agent's process. This means affected endpoints can appear to have a fully functional EDR agent running while the product is actually blind to new process, thread, and image-load activity. This distinction matters because it can create a false sense of assurance during incident triage if defenders rely solely on agent health status rather than validating telemetry continuity.
Organizations should view this campaign as a reminder that internet-facing IIS and Linux web servers are prime targets for initial access by attackers. Recommended mitigations include patching all publicly known vulnerabilities mentioned in this campaign, monitoring for BYOVD activities, and detecting kernel-level persistence mechanisms. Additionally, it is important to identify post-exploitation tools, as UAT-10147’s evolving tactics represent a growing intersection of AI-enabled operations and advanced malware capabilities.
In response to the release of this report, eSentire's Threat Intelligence team is actively tracking this topic for additional details and detection opportunities. IP addresses associated with this campaign have been added to the eSentire Global Block List, and the Threat Response Unit (TRU) is conducting threat hunts across customer environments.
CISA Adds Four Vulnerabilities to its KEV Catalog
Bottom Line: CISA added four actively exploited vulnerabilities to its Known Exploited Vulnerabilities catalog, critical flaws in Windows IKE Service Extensions, Microsoft SharePoint, Broadcom VMware vCenter, and Apple macOS Screen Sharing, with federal agencies required to remediate by August 21st, 2026.
On August 18th, 2026, The U.S. Cybersecurity and Infrastructure Security Agency (CISA) added four vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog. Per Binding Operational Directive (BOD) 26-04, Federal Civilian Executive Branch agencies must address these vulnerabilities by the specified due date, and CISA has ordered federal agencies to remediate all four flaws by August 21st, 2026.
CVE-2026-33824 (CVSS 9.8) - A critical Windows Internet Key Exchange (IKE) Service Extensions double-free vulnerability that could allow remote attackers to execute code on affected systems; while blocking UDP ports 500 and 4500 can reduce exposure to external attackers, internal actors could still exploit the flaw for lateral movement, making rapid patching the priority recommendation.
CVE-2026-55040 (CVSS 9.1) - A critical SharePoint authentication bypass allowing an unauthenticated attacker to exploit weaknesses in JWT validation to forge tokens and impersonate any SharePoint user, including administrators. Rapid7, which previously documented the flaw, described the root cause as a chain of four distinct weaknesses in SharePoint Server Subscription Edition's JWT validation pipeline that, combined, let an attacker forge a valid token without ever supplying a legitimate signature. Security researchers have reportedly already observed exploitation attempts against honeypot environments using a public Proof-of-Concept (PoC) for this vulnerability.
CVE-2026-59310 (CVSS 9.8) - A critical path traversal vulnerability in VMware vCenter's Syslog server that allows an attacker with network access to execute arbitrary code.
CVE-2026-65400 (CVSS: 7.1) - An attacker on the network may be able to authenticate to Screen Sharing without valid credentials. The Netherlands Cyber Security Centre (NCSC-NL) updated their advisory on August 12th, 2026, to state that public PoC exploit code is available and that the vulnerability is being actively exploited in the wild. Apple patched the issue in macOS Tahoe 26.6.1, macOS Sequoia 15.7.9, and macOS Sonoma 14.8.9.
eSentire Threat Intelligence Analysis:
Although CISA has added the four vulnerabilities to its KEV catalog, no information was shared regarding any observed attacks. However, third party reporting has previously identified the flaws being leveraged in real-world attacks, so it is unclear if CISA is basing their assessment off this information, or if additional campaigns were identified. Regarding CVE-2026-55040, the firm Defused observed exploitation within 24 hours of Rapid7’s PoC exploit script being released, highlighting how PoC can be rapidly adopted. For the CVE-2026-59310 flaw, security firm QUIRSO identified exploitation attempts linked to a suspected Chinese APT, leveraging the flaw to deploy Babuk ransomware. Finally, Microsoft identified exploitation of CVE-2026-65400 being used to transfer malicious files to impacted devices, ultimately deploying the XMRig cryptocurrency miner.
CISA provided the standard three-day deadline of applying relevant patches, as per BOD 26-04. The brief deadline has been linked to the use of AI to accelerate vulnerability exploitation, rapidly reducing the window from disclosure to attacks. Given the addition of the flaws to CISA's KEV catalog, and multiple reports of ongoing exploitation, organizations that utilize the impacted software within their environment should ensure that relevant patches are applied immediately. eSentire's Managed Vulnerability Service (MVS) has plugins in place to identify devices vulnerable to all four of the mentioned vulnerabilities, and eSentire's Threat Response Unit (TRU) is continuing to track the flaws for additional information and detection opportunities.
CISA Warns of Siemens S7 Series PLCs Being Targeted in Attacks
Bottom Line: U.S. government agencies issued a joint advisory warning of an active threat campaign targeting Siemens S7 Series Programmable Logic Controllers across water, energy, and other critical infrastructure sectors in the U.S.
On August 19th, 2026, CISA, the NSA, the FBI, the Department of Energy (DOE), and the Environmental Protection Agency (EPA) issued a joint advisory warning critical infrastructure organizations of attacks targeting Siemens S7 Series Programmable Logic Controller (PLC) devices. The agencies state that threat actors are scanning for Internet-exposed PLCs running “outdated software or that are otherwise poorly configured” and using AI-generated exploit scripts to conduct initial access, credential access, Denial of Service, and “other objectives”.
Successful attacks against PLCs could result in the disruption of critical industrial processes, safety incidents impacting personnel, equipment damage and downtime, the compromise of sensitive data, compliance violations, and supply chain impacts across interconnected systems. The advisory warns of attacks targeting critical infrastructure sectors within the U.S., specifically in the critical manufacturing, energy, water and wastewater, chemical, food and agriculture, and commercial facilities sectors. The agencies note that while the advisory focuses on an active threat to Siemens S7 Series PLCs, attacks targeting additional PLC brands are ongoing, and organizations should ensure that relevant mitigation steps are applied to reduce risk.
eSentire Threat Intelligence Analysis:
This latest advisory from CISA follows a July 2026 update to a previous advisory on Iranian APTs targeting PLC devices across U.S. critical infrastructure. In this previous advisory, CISA highlighted that attackers were able to “disable critical shutdown and alarm logic”, which would allow systems to enter unsafe conditions without proper notifications to operators. In between CISA's update and the recent advisory publication, several reports have emerged indicating cyber-attacks had impacted water systems in at least 12 U.S. states in July 2026. One of the attacks in Minnesota resulted in a municipal water system responsible for well and water treatment to be taken offline, and another in Georgia disrupted a county's water services. These reports demonstrate real-world impacts that these attacks can have. In the latest advisory, CISA does not provide any information on the threat actors behind the campaign, and at the time of writing, no formal attribution has been given to the attacks against water systems in the U.S.
The advisory highlights the use of AI-generated exploit scripts which were used to conduct attacks against the PLCs. Leveraging AI within attacks reduces the technical expertise and time required to develop exploits and other malicious tools, lowering the barrier of entry for threat actors to conduct large scale and high-impact campaigns. The use of AI within attacks can allow threat actors to leverage additional attack vectors and quickly adapt to defensive measures. To protect against this threat, organizations that utilize PLCs must perform an inventory of devices, ensure all security patches are applied, and implement network segmentation to ensure that PLCs are not accessible from the Internet. Default account names and credentials should be rotated, and strong passwords and Multi-Factor Authentication (MFA) should be enforced on all accounts where possible.
About the threat briefings:
The threat briefing is a weekly intelligence overview for non-technical users. The Threat Intelligence team at eSentire investigates, analyzes, and organizes the most important events of the past week along with important security tips and redistributes the findings for quick reading. The main goal of this briefing is to improve the overall awareness of readers regarding cyber security.
