It feels rare now to open any technology news site, or even any mainstream news site, without seeing headlines on a new cyber attack or data breach.
The number of reported Common Vulnerabilities and Exposures (CVEs) has more than doubled in 2026 from 2025, jumping from 48,244 records in 2025 to over 100,000 in 2026 to date (CVE, The Register).
Cybersecurity experts have warned that AI can be used to identify weaknesses in IT systems and scale up attacks. Based on this, businesses could well see an uptick in both the volume and impact of cyber threats.
In this insight, we explore how AI has changed cybersecurity, and how to protect your infrastructure and stay ahead of the curve.
How big is the challenge for cybersecurity?
The capabilities of AI technology mean developments in both cyber attacks and cyber defences are moving at a rapid rate.
Speaking on the impact of AI on security threats, Richard Horne, Head of the UK’s National Cyber Security Centre (NCSC), stated: “It can become a force multiplier increasing the speed and scale of our adversaries’ operations” (NCSC).
The UK government has also warned individuals and businesses throughout the country to prepare for supply chain disruption from emerging risks including cyber incidents. Speaking on Question Time, Defence Secretary Wes Streeting stated: “In this world of AI that we live in… it does also present threats if people are innovating and pushing against conventional [cyber] defences” (BBC).
Security researchers are using AI to accelerate code analysis, weakness identification, and vulnerability discovery. But the same capabilities can also be used by malicious actors to identify potential targets, automate elements of an attack and exploit weaknesses across multiple systems.
The good news is, so far, the rate of vulnerability discovery has increased at a faster rate than reported material breaches.
In a recent analysis, cybersecurity expert Tal Be’ery compared the growth in high- and critical-severity vulnerabilities with material cybersecurity incidents reported by U.S. public companies. While both increased during 2026, the rise in reported incidents was considerably smaller than the surge in vulnerabilities.
Analysis of the full picture, not just the vulnerability count, shows that while AI is increasing the opportunities for exploitation, other factors appear to be limiting the number of vulnerabilities that translate into successful attacks. Identifying and exploiting a vulnerability is only one stage of a successful attack. Attackers must also gain access to valuable systems or data and, in financially motivated attacks, find a way to profit from the compromise.
While these additional challenges may currently be limiting the number of vulnerabilities that translate into successful attacks, this balance could change as AI capabilities develop and attackers find new ways to monetise their activities.
This gap between vulnerability discovery and successful attacks strengthens the case for proactive security, patch management and ongoing infrastructure maintenance. As vulnerabilities are identified more quickly, businesses need to be able to assess their exposure and respond before attackers can exploit them.
Recent cyberattacks highlight the growing risks
Recent incidents demonstrate the different ways in which vulnerabilities and weaknesses can be exploited, from widely used software to connected infrastructure and internal AI tools. They also highlight the challenges businesses face in identifying risks and maintaining security across increasingly complex environments.
The PHP vulnerability
In October 2026, news broke of two critical vulnerabilities in PHP, a scripting language used across a large percentage of websites and applications. The vulnerabilities allow remote code execution, meaning attackers could potentially run commands on affected servers without user interaction or authentication.
The vulnerabilities were originally discovered by a group of cybersecurity researchers and organisations using frontier AI models. Cybersecurity firms Calif and Wiz reported the vulnerabilities, beginning a coordinated disclosure and patch development process. However, details were prematurely leaked online, creating an urgent need to protect affected sites before official patches were released.
The vulnerabilities affected sites and applications running vulnerable versions of PHP. Security agencies globally, including the UK’s National Cyber Security Centre (NCSC) and the US Cybersecurity and Infrastructure Security Agency (CISA), warned organisations to update their infrastructure immediately. Many sites, including UK government services (The Independent), were also taken offline while awaiting patches.
According to Matthew Prince, CEO of Cloudflare, the company recorded tens of thousands of exploitation attempts within just 45 minutes of the announcement, rising to hundreds of thousands within two hours and millions shortly afterwards (Matthew Prince, X).
Cold chain attacks across Europe
During October and November 2026, supermarkets across Europe were targeted by hackers who took control of refrigeration and freezer systems and caused temperatures to rise.
Blood banks in France were also targeted in an attempt to raise refrigerator temperatures and destroy blood plasma. In this instance, staff were alerted by a monitoring system and were able to prevent damage.
Investigations following the attacks indicated that some of the affected systems had reached end-of-life and were no longer receiving updates. Cybersecurity firm Claroty had previously identified vulnerabilities in the manufacturer’s control systems in 2025, which were subsequently patched. However, similar weaknesses remained present in other supermarket systems.
Although it has not been confirmed whether AI was used in these specific attacks, and investigations are ongoing, the incidents highlight how outdated software and poor security configurations can leave internet-connected devices and systems vulnerable. AI could make it easier for malicious actors to identify and exploit similar weaknesses in the future.
Binance cryptocurrency theft
Binance, the world’s largest cryptocurrency exchange, was hit by a sophisticated cyber attack, which began in June 2026, and culminated in the theft of £4.7bn in crypto-assets in October 2026.
The company reported that the attack began when an employee inadvertently downloaded malware which exposed details of the team responsible for Binance’s cryptowallet infrastructure. Another engineer was then targeted through a fraudulent recruitment process, consisting of two fake interviews, during the second of which the engineer was asked to download a coding assignment, which contained malware, onto their computer.
Once the engineer’s computer was compromised, attackers were able to leverage an internal AI tool to gain control of two engineers’ systems, and use compromised credentials and infrastructure permissions to deploy a malicious script, resulting in the theft.
The incident showed the security risks associated with interconnected systems and internal AI tools, particularly where access permissions and security controls could allow attackers to move between systems.
How to stay ahead of the threats
Understanding the risks of outdated and legacy systems
A common thread in many recent security incidents has been the exploitation of outdated and end-of-life systems. Once a system reaches end-of-life, it may no longer receive security patches and updates from the vendor, making newly discovered vulnerabilities more difficult to address.
Despite the security risks, many organisations continue to rely on legacy technology, often for understandable reasons. Replacing an older system isn’t always straightforward, particularly when it supports business-critical applications, needs to remain compatible with other technology, or would require significant investment to replace.
However, keeping these systems running without addressing the associated security risks now leaves businesses increasingly exposed.
For many years, vulnerabilities in these systems may have gone undiscovered or unexploited. The PHP2Shell incident demonstrated that AI is now making it possible to identify weaknesses at a speed and scale that would previously have been difficult to achieve. The vulnerabilities particularly affected sites running older versions of PHP, where applying security updates could be more challenging, further highlighting the risks associated with maintaining legacy software.
Some of these vulnerabilities are effectively hiding in plain sight. In the cold chain attacks, for example, default access passwords were reportedly published online in an operating manual, illustrating how relatively simple security weaknesses can leave systems exposed.
When weighing up whether to retain or replace legacy infrastructure, organisations should consider how the developing capabilities of AI could affect their exposure to cyber threats. While modernisation can seem like an unnecessary hassle, it is essential to take into account the full picture – the cost of upgrading a platform can easily be dwarfed by the consequences of an attack.
Where immediate replacement is not possible, additional security controls can help reduce the risks. These might include restricting access, implementing network segmentation, strengthening monitoring and isolating legacy infrastructure from other critical systems to limit the potential impact of an attack.
Patch management is key
As the rate of vulnerability discovery increases with the use of AI, so does the importance of proactive patch management. According to VulnCheck, the median time between CVE publication and exploitation fell from 120 days in 2025 to 80 days in the first half of 2026, meaning the time available to respond is shrinking (The Register).
Organisations need to be able to identify and address vulnerabilities promptly, before attackers have the opportunity to exploit them.
Effective patch management protects infrastructure by regularly monitoring for new vulnerabilities, assessing which systems are affected and prioritising updates based on the level of risk they present.
Additionally, as highlighted by the cold chain attacks, in some cases security weaknesses may already be known but have remained unaddressed. New discovery and attack methods mean that previous risk assessments may no longer be sufficient. In the face of these emerging threats, organisations should regularly reassess their infrastructure and address existing vulnerabilities, alongside monitoring for new vulnerabilities and patches.
However, timely patching must also be balanced with operational requirements. Updates need to be appropriately tested to ensure compatibility with existing systems and minimise the risk of disruption or downtime. This is particularly important for organisations running complex environments or business-critical applications, where an unsuccessful update could have significant consequences.
If your organisation uses a third-party provider for patch management, it is important that their engineers understand your infrastructure and its operational requirements. This knowledge helps them identify which vulnerabilities pose the greatest risk to your systems and determine how to address them without unnecessarily disrupting your operations.
Protecting your entire attack surface
An organisation’s attack surface covers every point an attacker could target, extending across servers, cloud infrastructure, applications, APIs, network devices, connected equipment and third-party services.
Over time, an organisation’s infrastructure can become increasingly complex. New applications are introduced, cloud environments are added, and different teams adopt their own tools and services. Without a coordinated approach to managing these systems, businesses can find themselves with piecemeal infrastructure and a lack of overall visibility.
This is often referred to as infrastructure sprawl, and it can create significant security challenges.
As infrastructure becomes more fragmented, the attack surface can expand, and security weaknesses are more likely to go unnoticed. Even in well-managed environments, the connections between systems can introduce risks. As demonstrated by the Binance attack, attackers can exploit weaknesses in interconnected systems and access controls to move between environments, potentially gaining access to mission-critical systems and sensitive data.
Maintaining a smaller, better-protected attack surface makes it more difficult for attackers to find entry points, helping businesses defend against emerging threats. Organisations should maintain a comprehensive inventory of their infrastructure and software to understand where potential weaknesses exist. Optimising infrastructure and bringing separate systems under centralised management can provide greater visibility and simplify security management, monitoring and updates.
Alongside improving visibility, organisations should ensure that appropriate security measures are implemented across their infrastructure. A multi-layered approach can include access controls, network segmentation, vulnerability scanning, endpoint detection and response, and continuous monitoring. AI-enabled security tools can also support threat detection and analysis, helping security teams identify suspicious activity and prioritise potential risks.
Together, these measures can help prevent unauthorised access, identify suspicious activity and limit the potential impact of a successful attack.
Proactive infrastructure management
In order to successfully stay ahead of threats, infrastructure management should be proactive, rather than just reactive. While security measures help protect against attacks, proactive management helps identify and address vulnerabilities before they become an attack vector.
Proactive management covers all aspects of an organisation’s infrastructure, and can include patch management, infrastructure modernisation, and layered security, depending on the specific provider’s offering.
A key part of this approach is ongoing monitoring and assessment. As new vulnerabilities are discovered and threats evolve, understanding what impacts your infrastructure is often the first challenge. From there, businesses need to assess the potential impact, decide how urgently it needs to be addressed and determine the safest way to implement any necessary changes.
This requires a detailed understanding of how your infrastructure works, including the applications it supports, how different systems interact and which services are critical to day-to-day operations.
Having experienced engineers who understand the environment makes this process more manageable. They can assess emerging risks in context, identify where action is needed and ensure that security updates and infrastructure changes are implemented with minimal disruption.
For organisations working with a managed hosting provider, this expertise can provide valuable support. Engineers who understand the customer’s environment can help identify where action is needed, manage updates and infrastructure changes, and provide ongoing oversight as new vulnerabilities emerge.
Your next steps: Staying ahead of evolving threats
The incidents explored in this article demonstrate how varied cybersecurity threats have become. From AI-assisted vulnerability discovery to attacks on connected refrigeration systems and the exploitation of internal AI tools, businesses are facing risks across an increasingly wide range of technologies.
While AI is introducing new capabilities for both attackers and defenders, many of the weaknesses being exploited are familiar. Outdated software, insecure configurations and compromised credentials have presented security challenges for years. However, as vulnerabilities can be identified and exploited more quickly, businesses may have less time to address these weaknesses before they become a target.
With the number of reported vulnerabilities continuing to grow, understanding which risks are relevant to your infrastructure is becoming increasingly important. Organisations need to know where their systems may be exposed, which weaknesses present the greatest risk and how to address them without compromising the performance and availability of business-critical services.
As AI continues to develop, businesses should be reviewing whether their existing approach to infrastructure management can keep pace. Proactive maintenance, ongoing security assessments and access to engineers who understand the environment can help organisations respond to new threats and maintain secure, reliable systems over the long term.





