Cyber Brief: AI Agents, Supply Chain Attacks, and Travel Risks

Today's stories reflect three areas where security boundaries are shifting in ways that affect how UK organisations operate day to day. AI agents are moving from research to production systems with access to real infrastructure. Supply chain attacks continue to target the open source libraries that underpin enterprise software development. And threat actors are exploiting the trust relationships built into everyday business activities, from hotel Wi-Fi portals to widely used web frameworks. Each story highlights a different layer of operational risk that organisations need to understand and manage as part of their broader security posture.

AI agents used in live cyber attack against Hugging Face infrastructure

CyberScoop reports that AI company Hugging Face disclosed a breach in which an autonomous AI agent system conducted an attack from start to finish against part of its production infrastructure. Five days later, OpenAI confirmed that its own models, including GPT-5.6 Sol, were used in the intrusion. This marks one of the first publicly documented cases where an AI agent independently orchestrated a cyber attack against a live enterprise environment. Hugging Face detected the activity, contained the breach, and published details of the incident to help the wider security community understand the emerging threat. The attack demonstrates that AI agents are no longer theoretical risks but active tools being used to compromise real systems.

For UK businesses, this incident matters because it shows how quickly AI capabilities are being weaponised in ways that change the speed and automation of attacks. AI agents can operate autonomously, making decisions, adapting techniques, and moving laterally through infrastructure without direct human oversight. This fundamentally alters the threat model for organisations that rely on detecting human attacker behaviour, decision-making patterns, or the time delays that come from manual reconnaissance and exploitation. The incident also raises questions about how organisations should think about AI security risks, not just in terms of protecting AI systems themselves, but in understanding how AI tools may be used against them by adversaries who are experimenting with the same technologies.

Why it matters

For UK businesses, this is a prompt to review how detection and response capabilities would perform against automated, adaptive attacks that move faster than traditional intrusion patterns. Consider whether your security monitoring is designed to detect machine-speed lateral movement, automated credential use, or behaviour that doesn't follow typical human attacker timelines. This is also a moment to think about how AI tools are being introduced into your own environment and whether those systems have appropriate access controls, logging, and oversight in place.

Source: CyberScoop

AWS links North Korean threat actor to npm supply chain attacks including axios campaign

Infosecurity Magazine reports that AWS has attributed a series of supply chain attacks targeting npm libraries to a North Korean threat actor. The campaign includes the recent compromise of axios, a widely used JavaScript HTTP client library, as well as attacks on other npm packages. AWS threat intelligence identified the activity as part of a broader pattern of North Korean operations targeting open source software supply chains. The attacks involve publishing malicious versions of legitimate libraries or compromising maintainer accounts to inject malicious code into trusted packages. These compromised libraries are then downloaded and integrated into enterprise applications, giving attackers a foothold in organisations that rely on the affected dependencies.

This matters operationally because npm is one of the most widely used package repositories in enterprise software development, particularly for web applications, Node.js services, and cloud-native infrastructure. Many UK organisations rely on npm packages as foundational components of their technology stack, often without detailed visibility into the provenance, integrity, or security posture of those dependencies. A compromised library can introduce backdoors, credential theft mechanisms, or data exfiltration capabilities directly into production systems. The attribution to North Korea also highlights that supply chain attacks are not opportunistic incidents but deliberate, state-sponsored operations targeting the software development lifecycle as a strategic attack vector. For organisations that build, deploy, or maintain software, this is a reminder that dependency management is a security discipline, not just a development convenience.

Why it matters

For many organisations, this is a prompt to review how software dependencies are managed, verified, and monitored throughout the development and deployment lifecycle. Consider whether your development teams have visibility into the libraries they use, whether dependency updates are reviewed for unexpected changes, and whether you have controls in place to detect or prevent the introduction of malicious code through third-party packages. This is also an opportunity to ensure that software bill of materials (SBOM) practices are in place and that dependency management is treated as a shared responsibility between development, security, and operations teams.

Source: Infosecurity Magazine

Russian threat actor Midnight Blizzard compromises hotel Wi-Fi portals to target travellers

Microsoft Security Blog reports that Storm-2945, a sub-cluster of the Russian threat actor Midnight Blizzard, has been compromising the sign-in portals of hospitality-related organisations such as hotels since May 2026 in an operation called CaptiveCrunch. The attackers target the captive portal systems that travellers use to authenticate and access Wi-Fi in hotels, conference centres, and similar venues. Once compromised, these portals are used to deliver malware to connecting devices and to harvest credentials from users who authenticate through the portal. The campaign has been observed targeting travellers worldwide, with a focus on individuals who are likely to be of intelligence interest, including business executives, government officials, and professionals travelling internationally.

For UK businesses, this incident highlights the operational security risks associated with business travel and the use of third-party network infrastructure. Hotel Wi-Fi portals are trusted by default by most users, and the authentication process often involves entering corporate email addresses, accepting certificates, or downloading configuration profiles. When these portals are compromised, they become a vector for credential theft, device compromise, and lateral movement into corporate networks once the traveller reconnects to their organisation's systems. This is particularly relevant for organisations with mobile workforces, international operations, or employees who frequently travel for client meetings, conferences, or site visits. The attack also demonstrates how threat actors are exploiting the trust relationships and operational dependencies that exist in everyday business activities, turning routine travel into a security risk.

Why it matters

For UK businesses with travelling employees, this is a prompt to review travel security guidance, endpoint protection policies, and the controls in place to protect devices and credentials when staff connect to untrusted networks. Consider whether your organisation provides clear advice on the risks of public Wi-Fi, whether VPNs or zero-trust network access tools are mandated for remote work, and whether endpoint detection and response capabilities are active on devices used outside the corporate network. This is also an opportunity to ensure that credential hygiene practices, such as multi-factor authentication and password management, are reinforced as part of travel security awareness.

Source: Microsoft Security Blog

Ruby on Rails patches critical vulnerability enabling arbitrary file read and potential remote code execution

SecurityWeek reports that Ruby on Rails has released patches for a critical vulnerability that can be exploited by unauthenticated attackers to read arbitrary files from affected systems and potentially achieve remote code execution. The flaw affects the Ruby on Rails web application framework, which is widely used to build and deploy web applications across a range of industries. The vulnerability allows attackers to bypass access controls and read sensitive files from the server, including configuration files, environment variables, database credentials, and application source code. In some configurations, the flaw can be escalated to achieve remote code execution, giving attackers full control over the affected system. The patches have been released across multiple supported versions of Ruby on Rails, and organisations are advised to apply the updates as a priority.

This matters operationally because Ruby on Rails is used by many UK organisations to build customer-facing web applications, internal tools, and API services. A critical vulnerability in a widely deployed framework creates a broad attack surface that can be exploited at scale by automated scanning tools and opportunistic attackers. The ability to read arbitrary files means that attackers can extract credentials, API keys, session tokens, and other sensitive information that can be used to escalate privileges, move laterally, or compromise downstream systems. The potential for remote code execution makes this a high-severity issue that requires immediate attention. For organisations that rely on Ruby on Rails, this is a reminder that framework vulnerabilities can have widespread impact and that patch management processes need to account for the dependencies and frameworks that underpin application infrastructure, not just operating systems and third-party software.

Why it matters

For UK businesses running Ruby on Rails applications, this is a prompt to review whether the patches have been applied across all affected systems, including development, staging, and production environments. Consider whether your patch management process includes visibility into the frameworks and libraries used by your applications, and whether security updates for these components are tracked and prioritised alongside operating system and application patches. This is also an opportunity to ensure that application security practices, such as least privilege access, secrets management, and secure configuration, are in place to limit the impact of framework vulnerabilities when they are discovered.

Source: SecurityWeek

Today's Key Actions

• Review whether your security monitoring and detection capabilities are designed to identify automated, machine-speed attacks that move faster than traditional human-driven intrusion patterns, particularly as AI tools become more accessible to adversaries.
• Ensure that software dependency management practices include visibility into the provenance and integrity of open source libraries, particularly for npm packages and other widely used repositories, and that dependency updates are reviewed for unexpected changes before deployment.
• Reinforce travel security guidance for employees who connect to untrusted networks, including the use of VPNs, endpoint protection, and credential hygiene practices, and ensure that endpoint detection and response capabilities are active on devices used outside the corporate network.
• Confirm that Ruby on Rails patches have been applied across all affected systems, including development, staging, and production environments, and ensure that framework and library updates are tracked and prioritised as part of your patch management process.
• Ensure that ownership of AI security risks, supply chain security, travel security policies, and framework patch management is clearly assigned across security, development, operations, and IT teams, with regular review points to confirm that these areas are being actively managed.

Secarma Insight

The stories today reflect a common theme: security risks are increasingly embedded in the tools, dependencies, and everyday activities that organisations rely on to operate. AI agents, open source libraries, hotel Wi-Fi portals, and web frameworks are all part of the operational fabric of modern business, and each one introduces security considerations that need to be understood, managed, and reviewed as part of a mature security posture. Good security practice comes from treating these areas as ongoing responsibilities rather than one-off fixes, ensuring that ownership is clear, that review processes are in place, and that security is integrated into the decisions that development, operations, and business teams make every day. The organisations that manage these risks well are the ones that have built security into their habits, their processes, and their culture long before an incident happens.