Cyber Brief: Critical BeyondTrust Flaws, AI-Assisted Ransomware

Today's brief focuses on vulnerabilities and attack techniques affecting widely deployed enterprise technologies. Two critical authentication bypass flaws in BeyondTrust's privileged access management tools require immediate attention, whilst emerging attack methods targeting AI agents and authentication flows demonstrate how threat actors are adapting to new technology patterns. These stories highlight the importance of timely patching, understanding how authentication mechanisms can be misused, and recognising that AI-assisted tooling is now part of the operational threat landscape for UK organisations.

Critical authentication bypass flaws in BeyondTrust privileged access tools

BeyondTrust has released patches for two critical vulnerabilities affecting its Remote Support (RS) and Privileged Remote Access (PRA) products, both widely used for managing privileged access in enterprise environments. The Hacker News reports that the flaws, tracked as CVE-2026-40138 (CVSS 9.2) and a second undisclosed critical issue, allow unauthenticated attackers to bypass authentication and potentially take control of affected systems. The vulnerabilities exist in pre-authentication code paths, meaning exploitation does not require valid credentials. BeyondTrust has published security advisories and patches, and organisations using these products are advised to apply updates immediately.

Privileged access management tools are high-value targets because they control access to critical systems, administrative credentials and sensitive infrastructure. A successful exploit could allow an attacker to move laterally across networks, escalate privileges, or establish persistent access to environments that should be tightly controlled. For UK organisations relying on BeyondTrust for remote support or privileged session management, these vulnerabilities represent a direct risk to the integrity of their access controls. The pre-authentication nature of the flaws makes them particularly serious, as they can be exploited without prior compromise of user accounts.

Why it matters

For UK businesses using BeyondTrust Remote Support or Privileged Remote Access, this is a prompt to review patch status urgently and ensure updates are applied as a priority. Organisations should also review access logs for any unusual authentication activity or unexpected privileged sessions, and confirm that privileged access management tools are included in routine vulnerability management and monitoring processes.

Source: The Hacker News

First documented AI-assisted ransomware attack shows evolving attacker capability

Sysdig has documented what it describes as the first known case of an AI agent being used to execute technical steps in a real-world ransomware attack, which took place in late June 2026. TechCrunch reports that whilst the AI agent handled much of the technical execution, including reconnaissance, lateral movement and deployment of ransomware payloads, a human operator still selected the victim organisation, set up the attack infrastructure, and provided stolen credentials. The attack demonstrates that AI agents can reduce the complexity and speed up the tempo of ransomware operations, but they are not yet fully autonomous. The threat actor used the AI agent to automate repetitive tasks and decision-making during the attack lifecycle, allowing faster progression through compromise stages.

This development reflects a broader trend of threat actors experimenting with AI tooling to improve operational efficiency and reduce the skill threshold required for complex attacks. For UK organisations, the operational implication is that ransomware attacks may become faster and more adaptive, with attackers able to respond dynamically to defensive measures or environmental conditions. The attack also underscores that stolen credentials remain a critical enabler, AI-assisted or not, the initial access vector still relied on compromised authentication. This reinforces the importance of credential hygiene, multi-factor authentication, and monitoring for unusual account activity as foundational defences.

Why it matters

For many organisations, this is a reminder that ransomware defences must account for faster attack progression and more adaptive adversary behaviour. UK businesses should review whether detection capabilities can identify rapid lateral movement, unusual automation patterns, or anomalous credential use, and ensure incident response plans account for the possibility of accelerated attack timelines.

Source: TechCrunch

Device code phishing attack exploits legitimate Microsoft authentication flow

Kaspersky researchers have identified a phishing campaign that abuses the OAuth 2.0 Device Authorization Grant, a legitimate authentication mechanism designed for devices such as Smart TVs and IoT hardware, to compromise Microsoft accounts. Securelist reports that attackers direct victims to genuine Microsoft websites and prompt them to enter a device code, which grants the attacker access to the victim's account without requiring traditional phishing infrastructure or credential capture. The attack is effective because the URL appears legitimate, the authentication flow is hosted by Microsoft, and users may not recognise the device code prompt as suspicious. Once the code is entered, the attacker gains persistent access to the account, often bypassing multi-factor authentication protections.

This technique highlights a broader challenge for UK organisations, authentication flows that are designed for convenience and device compatibility can be repurposed by attackers in ways that are difficult for users to detect. The attack relies on social engineering rather than technical exploitation, meaning traditional security controls such as email filtering or endpoint protection may not prevent it. For organisations using Microsoft 365, Azure AD, or other cloud identity platforms, this is a reminder that user awareness and conditional access policies are critical layers of defence. The attack also demonstrates that checking the URL alone is no longer sufficient to verify legitimacy, users need to understand the context of what they are being asked to do.

Why it matters

For UK businesses, this is a prompt to review user awareness training to include device code phishing scenarios, and to consider whether conditional access policies can restrict device code authentication flows to trusted locations or managed devices. Organisations should also review authentication logs for unexpected device registrations or unusual sign-in patterns that may indicate account compromise via this method.

Source: Securelist (Kaspersky)

Indirect prompt injection attacks target AI agents via web content

Zscaler researchers have identified a new attack technique targeting AI agents, in which malicious instructions are hidden within web content to manipulate the behaviour of AI systems that retrieve and process information from the internet. Infosecurity Magazine reports that attackers embed prompt-injection text into websites, which is invisible to human users but can be read and executed by AI agents. In documented cases, these hidden prompts instructed AI agents to initiate cryptocurrency payments or perform other unauthorised actions on behalf of the user. The technique exploits the way AI agents process external content, treating instructions found in web pages as legitimate commands rather than untrusted input.

This attack method is particularly relevant for UK organisations deploying AI agents for tasks such as research, content summarisation, customer support, or automated decision-making. If an AI agent retrieves information from a compromised or malicious website, it may execute instructions that were never intended by the user or the organisation. The risk is compounded by the fact that many AI agents operate with elevated permissions or access to sensitive systems, meaning a successful prompt injection could lead to data exfiltration, financial loss, or unauthorised system changes. For organisations experimenting with AI-driven automation, this is a reminder that AI agents require the same level of input validation, access control, and monitoring as any other privileged system component.

Why it matters

For UK businesses deploying AI agents, this is a prompt to review how these systems retrieve and process external content, and to ensure that AI agents operate with least-privilege access and appropriate guardrails. Organisations should also consider whether AI agent activity is logged and monitored in the same way as other automated systems, and whether security teams understand the risks associated with prompt injection and similar AI-specific attack techniques.

Source: Infosecurity Magazine

Today's Key Actions

• If your organisation uses BeyondTrust Remote Support or Privileged Remote Access, prioritise patching CVE-2026-40138 and related critical vulnerabilities immediately, and review access logs for any unusual privileged session activity.
• Review whether your detection capabilities can identify rapid lateral movement, unusual automation patterns, or anomalous credential use that may indicate AI-assisted or accelerated ransomware attacks.
• Update user awareness training to include device code phishing scenarios, and review conditional access policies to restrict device code authentication flows to trusted contexts where appropriate.
• If your organisation is deploying AI agents, review how these systems retrieve and process external content, ensure they operate with least-privilege access, and confirm that AI agent activity is logged and monitored appropriately.
• Ensure clear ownership and accountability for privileged access management, AI system governance, authentication policy, and emerging threat intelligence across your security and IT teams.

Secarma Insight

Today's stories reflect a consistent theme, attackers are adapting to the technologies organisations rely on, whether that's privileged access management tools, AI-assisted automation, or cloud authentication flows. Mature security practice means understanding that new capabilities bring new risks, and that defences must evolve alongside the tools we deploy. The organisations that manage these risks well are those that maintain clear ownership of emerging technologies, ensure security teams understand how new systems work, and build monitoring and response capabilities before incidents happen. Good security is not about reacting to every new threat with alarm, it's about maintaining the discipline, visibility and governance that allow you to respond confidently when risks emerge.