Cyber Brief: PHP Flaws, Supply Chain Attacks and AI Risk

Today's briefing is shaped by a common thread running through each of its four stories: the gap between how quickly threats emerge and how quickly organisations are able to respond. From a critical vulnerability affecting the majority of the web's server-side infrastructure, to a state-aligned supply chain attack using AI-assisted automation, to a twelve-year-old flaw in GitLab that has only now been fully patched, the picture is one of attackers moving fast and defenders needing clear, practical guidance to keep pace. For UK businesses, the message is consistent: patch promptly, understand your supply chain, and ensure that the tools your teams rely on are properly governed.

NCSC Urges UK Organisations to Patch Critical PHP Vulnerabilities Immediately

The National Cyber Security Centre has issued guidance urging UK organisations to take immediate action to address two critical vulnerabilities in PHP, tracked as CVE-2026-109618 and CVE-2026-109619 and collectively known as PHP2Shell. Reporting from the BBC, SecurityWeek, Krebs on Security and Wired confirms that these flaws allow unauthenticated remote code execution, meaning an attacker can take full control of an affected server without needing any credentials. PHP underpins approximately 70 per cent of websites globally, making the potential scale of impact significant. Patches are available in PHP versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35. SecurityWeek reports that exploitation attempts were detected by multiple security companies before patches were publicly released, and Wired confirms that Amazon Web Services logged active attempts four days prior to the patch release, with exploitation now described as broad and ongoing.

For UK businesses, the immediate concern is whether any web-facing infrastructure, applications or hosted services are running an unpatched version of PHP. This is not a theoretical risk: active exploitation is confirmed, and the vulnerability requires no authentication, meaning attackers do not need to have previously compromised an account or credential to gain access. Organisations that rely on managed hosting providers, content management systems such as WordPress, Drupal or Joomla, or custom web applications built on PHP should verify their patch status as a matter of priority. The NCSC's direct advisory to UK organisations underlines that this is a genuine and present concern rather than a precautionary notice.

Why it matters

For UK businesses running PHP-based web infrastructure, this is a prompt to verify patch status today and confirm with any managed hosting or third-party web service providers that updates have been applied. Organisations should also review whether web application firewall rules or other compensating controls are in place for any systems that cannot be patched immediately, and ensure that incident response contacts are available should signs of compromise emerge.

Source: NCSC UK

Microsoft Attributes ConnectWise Supply Chain Attack to China-Aligned Threat Group

Microsoft has published threat intelligence attributing a recent supply chain compromise of the ConnectWise Cloud Backup application to Storm-0940, a threat actor assessed to be aligned with Chinese state interests. Reporting from The Hacker News and CyberScoop details how Storm-0940 used a Go-based agentic AI penetration testing framework called ARTEX to automate the attack, reaching more than 2,250 customer tenants at a rate of approximately 2,900 sign-in attempts per minute. ConnectWise is widely used by managed service providers and IT teams for remote access, backup and endpoint management, making its customer base a high-value target for an actor seeking broad access to downstream organisations. The ARTEX framework had previously been made closed-source following its abuse against South Korean financial institutions, but Storm-0940 appears to have continued using it in this campaign.

This incident is directly relevant to UK organisations for two reasons. First, ConnectWise products are used extensively by UK managed service providers and their clients, meaning the potential blast radius of a supply chain compromise extends well beyond the immediate customer. Second, the use of AI-assisted automation to accelerate credential attacks at scale represents a meaningful shift in how sophisticated threat actors are operating. The speed at which Storm-0940 was able to probe tenants illustrates why traditional assumptions about the time available to detect and respond to an attack are being challenged. Organisations that rely on managed service providers should understand what tools those providers use and whether those tools have been affected.

Why it matters

For UK organisations using ConnectWise directly or through a managed service provider, this is a prompt to confirm whether any accounts or environments were affected and to review the access permissions granted to third-party management tools. More broadly, this is a useful moment to revisit supply chain security questions with key technology partners: what software do they use, how quickly do they patch, and how would they notify you in the event of a compromise affecting their tooling?

Source: The Hacker News

Critical GitLab Flaw Affects Every Release Going Back to 2014

GitLab released fixes on 21 October for a critical vulnerability carrying a CVSS score of 10.0 that allows an unauthenticated attacker to execute arbitrary code on self-managed GitLab servers. The Hacker News reports that the flaw affects every version of GitLab going back to version 7.0, released in 2014, meaning organisations running self-hosted GitLab instances across more than twelve years of releases are potentially exposed. Importantly, an earlier patch issued on 12 October only addressed versions 19.2 and later, leaving older installations unprotected until this week's update. Organisations that applied the earlier patch should verify whether the latest release is also required for their specific version.

GitLab is widely used by development and engineering teams for source code management, CI/CD pipelines and DevOps workflows. A successful exploit against a self-managed GitLab instance could give an attacker access to source code repositories, deployment credentials, pipeline configurations and potentially the ability to inject malicious code into software builds. For UK organisations in software development, technology services, financial services or any sector with active development teams, this represents a meaningful risk to both intellectual property and software supply chain integrity. The breadth of affected versions means that organisations running older or less frequently updated instances are particularly exposed.

Why it matters

For UK organisations running self-managed GitLab instances, this is a prompt to confirm that the latest patched version has been applied and that the earlier October patch alone is not considered sufficient. Teams should also review whether their GitLab instances are internet-facing and, if so, whether network-level access controls can reduce exposure while patching is completed. This is also a useful reminder to include developer tooling and source code management platforms within regular vulnerability management processes.

Source: The Hacker News

AI Agent Permissions Are Outpacing Security Governance

Help Net Security reports on new survey findings showing that 70 per cent of organisations have now granted AI agents the ability to make decisions and take actions that would previously have required human approval. The survey of 1,100 IT and security decision-makers also found that three in four organisations experienced at least one AI-related security issue or warning sign in the past twelve months. Separately, Infosecurity Magazine reports that researchers have uncovered a new phishing-as-a-service platform called Athena StealthRoute, which uses an integrated AI Studio to generate convincing phishing lures and decoy content designed to conceal malicious activity from detection tools. Together, these developments illustrate both the governance challenge of AI being adopted faster than security controls can keep pace, and the way attackers are using the same AI capabilities to improve the quality and scale of their own operations.

For UK organisations, the practical concern here is one of governance and oversight rather than technology alone. As AI tools are integrated into business workflows, they are increasingly being granted access to sensitive systems, data and decision-making processes without the same level of scrutiny that would apply to a human employee or a new software deployment. The combination of AI agents with broad permissions and AI-enhanced phishing attacks that are harder to detect creates a compounding risk: employees may be more likely to fall for a well-crafted AI-generated phishing lure, and the consequences of a successful attack may be amplified if AI agents with elevated access have already been compromised or manipulated.

Why it matters

For UK businesses deploying AI tools, this is a prompt to review what permissions and system access have been granted to AI agents and whether those permissions follow a least-privilege principle. Security and IT teams should also ensure that AI-enhanced phishing is reflected in current awareness training, so that employees understand that AI-generated content may be more convincing than traditional phishing attempts. Governance frameworks for AI adoption should include security sign-off as a standard step before new tools are granted access to sensitive systems or data.

Source: Help Net Security

Today's Key Actions

• Verify that all PHP installations across your web infrastructure have been updated to a patched version (8.5.12, 8.4.27, 8.3.36 or 8.2.35) and confirm patch status with any managed hosting or third-party web service providers.
• If your organisation uses ConnectWise directly or through a managed service provider, confirm whether any accounts or environments were affected by the supply chain compromise and review the access permissions granted to third-party management and remote access tools.
• Check that all self-managed GitLab instances have been updated to the latest patched release and that the October 12 patch alone has not been treated as sufficient; confirm whether any instances are internet-facing and apply network-level access controls where patching cannot be completed immediately.
• Review the permissions and system access granted to AI agents and tools within your organisation, apply least-privilege principles, and ensure that AI-enhanced phishing techniques are reflected in current security awareness training.
• Confirm that clear ownership exists for each of these areas — web infrastructure patching, third-party tool governance, developer platform security and AI adoption oversight — so that follow-up actions can be assigned and tracked effectively.

Secarma Insight

Today's stories are a useful reminder that effective security is rarely about responding to one dramatic event in isolation — it is about having the processes, ownership and habits already in place that allow an organisation to respond quickly and confidently when something significant does emerge. The PHP vulnerability, the ConnectWise supply chain attack and the GitLab flaw all share a common characteristic: organisations with clear patch management processes, well-governed third-party relationships and regular vulnerability reviews will be in a far stronger position than those responding from scratch. The AI governance story adds a further dimension, reflecting that the same discipline of clear ownership and proportionate controls applies just as much to emerging technology adoption as it does to traditional infrastructure. Mature security practice is not about eliminating all risk — it is about knowing where your exposure lies and having the confidence to act on it.