Anthropic announced that its internal AI models under “high-risk and cybersecurity evaluations” no longer have internet access following incidents involving U.S. government websites at the federal, state, and local levels. The instances leading to the decision include AI models that exploited website vulnerabilities, submitted forms, bypassed data restrictions, and circumvented safeguards. However, the company said that the misaligned behavior had a minimal real-world impact and assessed the behavior to be less severe than previously reported incidents. The AI company notified the White House about the unauthorized activity: Anthropic estimates that the agents may have been motivated by overly generous rewards during reinforcement learning, which may have encouraged them to find unauthorized workarounds rather than stop when a task was blocked or impossible.
CyberXero used 51 AI agents to support cyberattacks, stealing 628,000 Ukrainian records, including sensitive data from energy infrastructure. CyberXero is an operation that combines traditional hacking tools with AI agents to carry out cyberattacks. According to SOCRadar researchers, the attackers use the same infrastructure for two different campaigns: one targets WordPress websites and e-commerce platforms worldwide, while the other focuses on Ukrainian energy and critical infrastructure. The actor behind CyberXero is a Russian-speaking, financially motivated threat actor. This type of attacker, known as an initial access broker, breaks into networks and sells access to other criminals instead of using it directly. One campaign focuses on attacking as many targets as possible, while the other involves carefully selected targets in Ukraine. “SOCRadar’s Threat Research Unit (STRU) has documented CyberXero, a Russian-speaking, financially motivated Initial Access Broker that pairs commodity offensive tooling with an AI orchestration layer running on its own infrastructure.” reads the report published by SOCRadar. “The operation runs two pipelines at once: a global, automated campaign against WordPress and e-commerce platforms, and a curated, manual campaign against Ukrainian energy and critical infrastructure. More than 628,000 Ukrainian individuals have confirmed data in the actor’s possession, including residents of Kharkiv, a city on an active war front.” The investigation only happened because the threat actor made one classic mistake, an open directory with zero access control sitting on a server. That single misconfiguration exposed the actor’s entire working directory, more than 90,000 files, including AI session logs, scripts with plaintext tokens, and already-stolen victim data. The exposed server was found at “46.21.250[.]135,” hosting a live Linux workstation’s complete contents. Session logs recording every planned target, every refused task, and every workaround attempt are, in practice, the offensive equivalent of a keylogger running against the attacker’s own operation. The global pipeline runs on a custom tool called wp2shell, which abuses a flaw in the WordPress REST API to inject SQL, plant a rogue admin account, and drop a webshell. The scale is genuinely industrial. A single automated run scanned 4,708 targets, confirmed access to 429 WordPress admin panels, and deployed 32 webshells, all within 61 seconds. That’s not a person clicking through targets one at a time, that’s infrastructure built to move at machine speed. The target list for this side of the operation exceeds 56,000 identified URLs, spanning e-commerce platforms and WordPress sites with no particular country or sector pattern. The second campaign is more targeted. The attacker manually selected targets in Ukraine, focusing on energy and utility companies, including the country’s national electricity transmission operator. Researchers confirmed data theft from four Ukrainian organizations. One of the most serious cases involved a district heating provider in Kharkiv, a city close to the front line. A password stored directly in the attacker’s script gave access to the provider’s systems. An automated tool then extracted more than 564,000 customer records, including home addresses and account numbers. SOCRadar warns that this data could put people at physical risk because Kharkiv is in an active war zone. Stolen customer information is usually a privacy concern, but in this situation, linking real names to home addresses could create much more serious dangers. “Because Kharkiv sits on an active conflict front, this dataset poses a potential physical risk to identifiable civilians that goes well beyond a conventional data breach.” continues the report. “The most elaborate chain, against an e-commerce portal, combined blind SQL injection, a time-constrained password-reset token extraction, and PHP deserialization to reach remote code execution and full database access.” The attacks stand out from typical cyber operations because the actor ran up to 51 specialized Claude Code agents on the primary workstation, assigned to roles like exploit chaining, lateral movement, and reconnaissance, plus a separate AI pentesting framework wired directly into a Cobalt Strike command server. SOCRadar’s researchers found the actor had also built a reusable script to get around model safety refusals, framing every request as an authorized penetration test on infrastructure the actor supposedly owned. That trick isn’t a technical bypass, it’s social engineering aimed at the model instead of a person. And it didn’t always work. The logs show the model declining real requests multiple times, including four separate refusals to install a backdoor, disable a firewall, move laterally, or deploy a webshell. Safety measures held in plenty of individual sessions. The actor’s workaround was just opening a new session each time, since a refusal doesn’t carry forward once the context resets. Researchers identified the operator by combining several clues: a paid AI subscription linked to a personal email address, a forum post using the same online alias, and activity that consistently matched Moscow time. The campaign affected more than 40 organizations across Ukraine, Poland, China and Pakistan. The main lesson for defenders is to protect AI agent session logs as carefully as passwords and other sensitive credentials. These logs can reveal an attacker’s plans, the systems they accessed, what worked and where their attempts failed. If exposed, they can provide a detailed picture of the entire operation. The report includes Indicators of Compromise (IoCs) for this campaign along with detection and hunting guidance. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, CyberXero)
A maximum-severity security flaw impacting the Bricksforge plugin for WordPress has come under active exploitation in the wild, according to Patchstack. The critical vulnerability in question is CVE-2026-85097 (CVSS score: 10.0), which allows unauthenticated attackers to upload and execute arbitrary PHP code on vulnerable WordPress websites, resulting in remote code execution (RCE). The
Cybersecurity researchers have disclosed details of a two-pronged artificial intelligence (AI)-augmented initial access campaign targeting WordPress sites, e-commerce platforms, and critical infrastructure entities in Ukraine. The activity, which brings together commodity offensive tools like Cobalt Strike with an AI orchestration layer powered by Anthropic's Claude Code, has been attributed
U.S. Cybersecurity and Infrastructure Security Agency (CISA) adds ProFTPD, ONLYOFFICE Docs, Strapi, Apache Struts, and ISC BIND to its Known Exploited Vulnerabilities catalog. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) added the following vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog: CVE-2015-3306 (CVSS score of 10.0) – An access control flaw in ProFTPD that could let remote attackers read or modify arbitrary files by abusing the SITE CPFR and SITE CPTO commands. CVE-2021-3199 (CVSS score of 9.8) – A path traversal flaw in ONLYOFFICE Docs when JSON Web Token (JWT) is enabled. Attackers could exploit a /.. sequence in an image upload parameter to potentially execute code remotely. CVE-2023-22894 (CVSS score of 7.2) – A vulnerability in Strapi that exposes sensitive information stored in cleartext. An attacker with admin panel access could use query filters to retrieve confidential user details. CVE-2016-3081 (CVSS score of 8.1) – A command injection flaw in Apache Struts that could allow remote attackers to run arbitrary code through method: prefixes when Dynamic Method Invocation is enabled. CVE-2015-5477 (CVSS score of 7.5) – A reachable assertion vulnerability in ISC BIND that could be triggered by remote TKEY queries, potentially causing a denial-of-service condition. The five vulnerabilities have been added to a broader list of flaws linked to cyber operations attributed to China-linked actors associated with Integrity Technology Group, a China-based cybersecurity company. The update coincides with a joint advisory issued by Australia, Canada, Japan, New Zealand, Spain, the United Kingdom and the United States. U.S. authorities have also taken action against tools associated with Integrity Tech and used in cyber espionage operations. The activity reportedly involved the exploitation of eight vulnerabilities, including the five listed above, to gain initial access to targeted networks and steal sensitive information. The attackers used scanning tools, cross-site scripting (XSS) and password-spraying attacks against Microsoft Exchange servers. They also relied on VPN software to maintain access and scripts to extract emails and credentials. The campaign is part of a broader set of activities linked to Integrity Tech. The U.S. Department of Justice and FBI seized two tools, Microscan and FishHub, allegedly operated by the Chinese company. Microscan was used to scan networks for vulnerable systems, while FishHub relied on spear-phishing emails to deliver malware, enable remote access and steal files. The tools were reportedly used against critical infrastructure and other organizations in multiple countries. The joint advisory highlights the risks posed by tools that combine large-scale vulnerability scanning with hands-on exploitation. The coordinated action by seven countries and the U.S. seizure of Microscan and FishHub aim to disrupt infrastructure allegedly used to support China-linked cyber operations. According to Binding Operational Directive (BOD) 22-01: Reducing the Significant Risk of Known Exploited Vulnerabilities, FCEB agencies have to address the identified vulnerabilities by the due date to protect their networks against attacks exploiting the flaws in the catalog. Experts also recommend that private organizations review the Catalog and address the vulnerabilities in their infrastructure. CISA orders federal agencies to fix the flaws by October 11, 2026. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, CISA)
Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock (LockA locked padlock) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites. Share: Opens in a new window Opens in a new window Opens in a new window Press Release CISA, FBI, NSA and International Partners Warn of China-based Cybersecurity Company Enabling Threat Actors to Target Multiple Critical Infrastructure Sectors Worldwide Advisory Provides Actionable Steps to Detect, Respond to, Reduce the Threat of Compromise to Malicious Activity Released October 08, 2026 Related topics: Cyber Threats and Response , WASHINGTON – Today the Cybersecurity and Infrastructure Security Agency (CISA), Federal Bureau of Investigation (FBI), National Security Agency (NSA), and international partners issued a joint Cybersecurity Advisory warning that Integrity Technology Group (Integrity Tech), a China-based cybersecurity company, is enabling threat actors to target multiple critical infrastructure sectors worldwide using a range of sophisticated tools, such as large-scale botnets, virtual private network infrastructure and living off the land. Based on real-world investigations and observed activity in North America, Southeast Asia and Africa, this advisory provides recommended mitigations and information to help network defenders detect and respond to this malicious activity.
Cybersecurity researchers have discovered a cluster of 16 malicious Mozilla Firefox extensions that are capable of stealing cryptocurrency wallet recovery phrases and private keys. "The extensions masquerade as wallet portals, desktop utilities, and browser tools, but their code intercepts recovery phrases and private keys during wallet import flows and attempts to send those secrets to
Unit 42 details how threat actors leverage Web3 infrastructure and open-source supply chain attacks to breach enterprise cloud environments The post Evolution of Web3 in Cloud Supply Chain Attacks appeared first on Unit 42.
GitLab fixes critical AI Gateway flaw that could let authenticated Duo users escape a prompt sandbox and execute commands on self-hosted gateways. GitLab has released patches for a critical vulnerability in its AI Gateway, tracked as CVE-2026-90970 (CVSS score of 9.9), that could allow an authenticated user with access to the Duo Agent Platform to execute arbitrary commands on the gateway. GitLab disclosed the flaw on October 2 and fixed it with the release of AI Gateway versions 19.2.4, 19.3.2 and 19.4.1. The issue affects the way the AI Gateway handles custom flow prompt templates. According to GitLab, a user with Duo Agent Platform access could use a specially crafted flow configuration to escape the prompt template sandbox and execute commands on the AI Gateway host. “GitLab has remediated an issue in the GitLab AI Gateway that, under certain conditions, could have allowed an authenticated user with Duo Agent Platform access to escape the prompt template sandbox via a specially crafted flow configuration, leading to arbitrary command execution on the AI Gateway.” reads the advisory. The important point is that the attack does not require an unauthenticated connection. GitLab’s advisory does not provide further details about the exact conditions required to exploit the flaw or the permissions needed beyond that access. GitLab credited the HackerOne researcher invisiblemeerkat with responsibly reporting the vulnerability. The GitLab AI Gateway is basically the middle layer between GitLab Duo and the AI models. GitLab Duo provides AI features such as code suggestions, chat, code explanation and agentic workflows. The AI Gateway receives the requests from GitLab, prepares the prompts and communicates with the underlying LLMs. GitLab operates its own cloud-based gateway, but organizations can also deploy an AI Gateway inside their own infrastructure. GitLab says the security fix has already been deployed to its own hosted AI Gateways. Customers using GitLab.com, GitLab Dedicated, or a GitLab Self-Managed instance connected to a GitLab-hosted gateway therefore do not need to take action. The customers that need to patch are those running a GitLab Self-Hosted AI Gateway. GitLab has already contacted those customers directly and strongly recommends updating affected installations immediately. Self-hosting the gateway is designed for organizations that want to keep AI requests and responses inside their own environment. In a fully self-hosted configuration, GitLab, the AI Gateway and the selected AI models can all operate within the organization’s infrastructure. Below are the affected AI Gateway releases: Affected versionFixed version18.1.6 through 19.2.319.2.419.3.0 through 19.3.119.3.219.4.019.4.1 The AI Gateway sits between GitLab’s AI features and the underlying models. In a self-hosted deployment, it processes requests between the GitLab instance and the organization’s AI infrastructure. The service also has access to sensitive authentication material. GitLab’s installation documentation says that self-hosted deployments require signing and validation keys for JSON Web Tokens. These keys are passed to the gateway as environment variables. For that reason, command execution on the gateway could provide an attacker with a foothold inside infrastructure that handles AI requests and authentication. The exact impacts would depend on how the gateway is deployed and what access it has to the rest of the organization’s environment. GitLab’s October 2 advisory does not say that CVE-2026-90970 has been exploited in the wild. It also does not publish a proof of concept or technical exploitation steps. That leaves several important questions unanswered, including the precise flow configuration needed to escape the template sandbox and the exact conditions under which the resulting command execution can be reached. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, newsletter)
GitLab warned customers today to immediately patch a critical AI Gateway vulnerability that could let attackers run arbitrary commands on vulnerable instances. AI Gateway is a service that gives access to AI-native GitLab Duo features. While GitLab operates its own cloud-based AI Gateway instance used by GitLab.com, GitLab Self-Managed, and GitLab Dedicated, users can also deploy their own self-hosted instances on GitLab Self-Managed through GitLab Duo Self-Hosted. Tracked as CVE-2026-90970, this security flaw stems from an improper neutralization weakness and can let attackers with basic privileges and Duo Agent Platform access execute arbitrary commands on unpatched instances. "GitLab has remediated an issue in the GitLab AI Gateway that, under certain conditions, could have allowed an authenticated user with Duo Agent Platform access to escape the prompt template sandbox via a specially crafted flow configuration, leading to arbitrary command execution on the AI Gateway," the company explained in a Friday advisory.
Confirm this action.
Leaving now will discard your changes.