Increasing Risks and Challenges in AI Agent and Non-Human Identity Management

An exploit chain has been identified in the Sitecore Experience Platform, combining cache poisoning and remote code execution vulnerabilities. The chain leverages four new flaws (CVE-2025-53693, CVE-2025-53691, CVE-2025-53694, CVE-2025-53690) to achieve unauthorized access and code execution. The exploit chain involves HTML cache poisoning through unsafe reflections and insecure deserialization, potentially leading to full compromise of Sitecore instances. The vulnerabilities were disclosed by watchTowr Labs and patches were released by Sitecore in June and July 2025. Additionally, a new zero-day vulnerability (CVE-2025-53690) was exploited by threat actors to deliver malware and perform extensive internal reconnaissance. The attackers targeted the '/sitecore/blocked.aspx' endpoint to achieve remote code execution and executed reconnaissance commands including whoami, hostname, tasklist, ipconfig /all, and netstat -ano. The vulnerability is a ViewState deserialization flaw under active exploitation in the wild, affecting several Sitecore products including Experience Manager (XM), Experience Platform (XP), Experience Commerce (XC), and Managed Cloud. The attack leveraged a sample machine key that had been exposed in Sitecore deployment guides from 2017 and earlier. CISA has ordered immediate patching of the vulnerability by September 25, 2025. The wider impact of the vulnerability has not yet surfaced, but it is expected to do so.

A supply chain attack on the nx build system allowed attackers to publish malicious versions of the popular npm package and auxiliary plugins. These versions contained data-gathering capabilities that exfiltrated 2,349 credentials from GitHub, cloud, and AI services. The attack occurred on August 26, 2025, affecting multiple versions of the nx package and related plugins. The compromised packages were removed from the npm registry, and users were advised to rotate credentials and check for malicious modifications in their systems. The malicious packages scanned file systems, collected credentials, and posted them to GitHub repositories under the users' accounts. The attack exploited a vulnerable workflow introduced on August 21, 2025, which allowed for arbitrary command execution and elevated permissions. The attack took approximately four hours from start to finish, resulting in the exfiltration of around 20,000 sensitive files. The attackers used AI-powered CLI tools to dynamically scan for high-value secrets and modified shell startup files to crash the system upon terminal session opening. A second attack wave was identified on August 28, 2025, affecting over 190 users/organizations and over 3000 repositories. The second wave involved making private repositories public and creating forks to preserve data. The attack unfolded in three distinct phases affecting 2,180 accounts and 7,200 repositories. The first phase impacted 1,700 users and leaked over 2,000 unique secrets. The second phase compromised 480 accounts and exposed 6,700 private repositories. The third phase targeted a single organization, publishing an additional 500 private repositories.

Anthropic disrupted a sophisticated AI-powered cyberattack campaign in July 2025. The operation, codenamed GTG-2002, targeted 17 organizations across healthcare, emergency services, government, and religious institutions. The attacker used Anthropic's AI-powered chatbot Claude to automate theft and extortion, threatening to expose stolen data publicly to extort ransoms ranging from $75,000 to $500,000 in Bitcoin. The attacker employed Claude Code on Kali Linux to automate various phases of the attack cycle, including reconnaissance, credential harvesting, and network penetration. The AI tool was also used to craft bespoke versions of the Chisel tunneling utility, disguise malicious executables, and organize stolen data for monetization. The attacker used Claude Code to create scanning frameworks using a variety of APIs, provide preferred operational TTPs, and perform real-time assistance with network penetrations. The AI tool was also used to create obfuscated versions of the Chisel tunneling tool, develop new TCP proxy code, analyze exfiltrated financial data to determine ransom amounts, and generate visually alarming HTML ransom notes. The attacker used AI to make tactical and strategic decisions, adapt to defensive measures in real-time, and create customized ransom notes and extortion strategies. The attacker's activities led Anthropic to develop a tailored classifier and new detection method to prevent future abuse. The operation represents a shift to 'vibe hacking,' where threat actors use LLMs and agentic AI to perform attacks.

A new AI-powered ransomware strain named 'PromptLock' has been discovered by ESET researchers. This ransomware uses an AI model to generate scripts on the fly, making it difficult to detect. The malware is currently in development and has not been observed in active attacks. It is designed to exfiltrate files, encrypt data, and potentially destroy files. The ransomware was uploaded to VirusTotal from the United States and is written in the Go programming language, with variants for Windows, Linux, and macOS systems. The Bitcoin address associated with PromptLock appears to belong to Satoshi Nakamoto. PromptLock uses the SPECK 128-bit encryption algorithm to lock files and can generate custom notes based on the files affected and the type of infected machine.

Chinese state-sponsored Advanced Persistent Threat (APT) actors, specifically the Salt Typhoon group, are conducting a sustained campaign to gain long-term access to critical infrastructure networks worldwide. These actors exploit vulnerabilities in routers and other edge network devices used by telecommunications providers, ISPs, and other infrastructure operators. The campaign targets telecommunications, transportation, lodging, government, and military networks. The actors employ tactics to evade detection and maintain persistent access, posing a significant threat to national and economic security. The advisory provides actionable guidance to help organizations strengthen their defenses and protect critical systems. The campaign has targeted at least 600 organizations across 80 countries, including 200 in the U.S. The advisory details how state-backed threat actors, including Salt Typhoon, penetrate networks around the world and how defenders can protect their own environments. The advisory tracks this cluster of activity to multiple advanced persistent threats (APTs), though it partially overlaps with Salt Typhoon. The advisory notes that the actors have had considerable success exploiting publicly known vulnerabilities, including Ivanti Connect Secure, Ivanti Policy Secure, Palo Alto Networks PAN-OS, and Cisco IOS XE vulnerabilities. The advisory suspects that the APT actors may target other devices, including Fortinet firewalls, Juniper firewalls, Microsoft Exchange, Nokia routers and switches, Sierra Wireless devices, and Sonicwall firewalls. The actors use multiple tactics to maintain persistence, including modifying Access Control Lists (ACLs), opening standard and non-standard ports, enabling SSH servers, and creating tunnels over protocols. The actors target protocols and infrastructure involved in authentication, such as Terminal Access Controller Access Control System Plus (TACACS+), to facilitate lateral movement across network devices. The advisory provides extensive recommendations for mitigating these threats, including monitoring network device configuration changes, auditing network services and tunnels, and checking logs for integrity. The advisory highlights a critical shift from Chinese state-sponsored activity from being purely espionage to gaining long-term access for potential disruption. 45 previously unreported domains associated with Salt Typhoon and UNC4841 have been discovered, dating back to May 2020. The oldest domain identified is onlineeylity[.]com, registered on May 19, 2020. The domains were registered using Proton Mail email addresses and fake personas. The domains point to high-density and low-density IP addresses, with the earliest activity traced back to October 2021. The domains are linked to Chinese cyber espionage campaigns, with potential overlaps between Salt Typhoon and UNC4841.