AI Storage

Prepare for heightened AI-powered cyberattacks with the right enterprise storage solution

a spiral notebook with the letter a on it

Written By Eric Herzog

Cybercriminals are evolving their attack methods and increasingly using artificial intelligence to carry out attacks that are faster, more advanced, and harder to detect. Tools once used to create deepfakes of celebrities for social media are being applied in enterprises to fake CEO communications and wreak havoc across existing cyber defences. For instance, AI-generated phishing can convincingly impersonate trusted contacts. But this is just the tip of the iceberg. Automated tools can rapidly scan systems for vulnerabilities, and AI-powered malware can adapt its behaviour to evade conventional security tools. These are not risks for tomorrow, but real and current enterprise threats detailed by the UK’s Information Commissioner’s Office in 2026. Generative AI has now become so sophisticated and readily available, it has brought renewed urgency and a greater requirement for speedy action against the, already, exceptionally high threat of cybercrime.

 

The UK’s National Cyber Security Centre (NCSC) verifies this advice and also warns about the role of AI in changing the cyber-threat landscape. It is making existing attack techniques more effective and efficient, rather than creating entirely new forms of cyberattack. The NCSC’s assessment of likely threats between now and 2027 identifies AI-assisted reconnaissance, vulnerability research and exploitation, social engineering and malware development as areas where AI can enhance attackers’ capabilities.

 

The significance for enterprises is not simply that individual attacks are becoming more sophisticated – this is only part of the problem. AI has the potential to accelerate parts of the attack lifecycle too. In particular, the NCSC expects AI-assisted vulnerability research and exploit development to compress the time between vulnerabilities being disclosed and then exploited by threat actors. This means enterprises need storage that is fully equipped for today’s AI-powered cyberattacks.

 

The new AI cybercriminal’s arsenal

The ways in which AI can be applied to cybercrime are already extensive. For example:

 

  • Dark LLMs are purpose-built, “uncensored” AI models that threat actors can use to write polymorphic malware and phishing scripts without safety guardrails. These types of outputs can change code or appearance as they spread or execute, while retaining the same harmful purpose.

 

  • Deepfake-as-a-Service uses synthetic video and audio cloning tools to impersonate corporate executives or other trusted individuals.

 

  • Automated reconnaissance enables AI bots to scan thousands of network systems for security misconfigurations and analyse publicly available information to create highly persuasive, targeted scams.

 

  • Just-in-time malware can use connected AI models to generate malicious functions and obfuscate code on demand, helping it evade traditional antivirus software.

 

The speed at which all this is possible compounds the problem. A recent research report by Booz Allen Hamilton[1] reports cybercriminals moving from initial access to broader system compromise in less than 30 minutes on average – and sometimes in seconds. AI is so powerful, it is even helping cybercriminals create these realistic phishing emails, research multiple targets rapidly and write malicious code without the need for existing coding skills. It can also enable smaller groups to carry out campaigns that would previously have required larger, coordinated groups. For enterprises, this creates an important change in the risk equation: much faster and more scalable attacks and less time in which to respond. AI has effectively lowered the technical barriers to some forms of cyber activity but at the same time, enabled technically proficient threat actors to operate more efficiently and at greater scale.

 

Changing role of enterprise storage

At the same time, expectations of enterprise storage have been evolving. Although capacity and performance remain essential, they are no longer the sole measures of success. Storage platforms are increasingly expected to simplify operations, support a wider range of workloads and help IT organisations respond to growing business demands without adding unnecessary complexity or cost. Enterprise cyber storage capabilities have become an essential requirement and a key factor in the transition for enterprise storage from being a “backroom capability” to a highly strategic enterprise asset.

 

Cyber resilience is an important part of this transition. More than 90% of an enterprise’s data sits in its enterprise storage infrastructure. Not having that data protected from a cyber-attack or securely ring-fenced to investigate the implications of an attack, is a major security oversight. The only feasible response is a “cyber storage resilient, recovery-first strategy” designed to address ransomware, malware and other cyber threats.

 

Recovery first approach

If AI-enabled cyberattacks are inevitable, enterprises need to be prepared with a recovery first survival strategy. At the same time, it is important to ensure that the data being recovered is a known good copy and free from embedded malware or ransomware. This can be difficult to achieve because malware, ransomware, and other cyberattacks are designed to operate surreptitiously. Without the right technology in place, it can therefore be difficult to establish whether malware or ransomware is already present within enterprise storage.

 

A fenced forensic environment has an important role here. Once data is inside the fenced forensic environment, application owners can verify that nothing has been damaged and establish that they have a known good copy to recover.

 

Enterprise storage also needs to work with the wider cybersecurity infrastructure. When reviewing enterprise storage solutions, enterprises should identify whether they have the ability to interface between SIEM and SOAR data-centre-wide cybersecurity applications and their storage infrastructure. If Microsoft Sentinel or the Security Operations Centre detects a cyberattack, there is no need to wait: immutable snapshots can start to be taken instantaneously.

 

These capabilities reflect how far the role of enterprise storage has evolved. Modern enterprise storage is no longer simply about storing data. It has the potential to drive cyber resilience and cyber recovery back into the enterprise, reducing the risks of disruption and protecting the data infrastructure.

 

What is coming in 2027?

Looking towards the immediate future, the NCSC expects AI models to become even more capable and increasingly accessible. AI is expected to continue making cyber intrusion more effective and efficient, with more advanced systems providing greater assistance with vulnerability discovery and exploit development. This could increase the speed with which attackers take advantage of weaknesses. Importantly, this does not mean autonomous AI systems suddenly replacing human cyber attackers. Human expertise is expected to remain important, particularly for sophisticated operations, while AI increasingly acts as an accelerator and force multiplier.

 

For enterprise storage, the direction of travel is equally clear. Enterprises need to be able to recover rapidly from a cyberattack while ensuring that the data being recovered is a known good copy. Enterprise storage infrastructure needs to work with wider cybersecurity applications, while fenced forensic environments and immutable snapshots can form part of a cyber storage resilient, recovery-first strategy.

 

Modern enterprise storage is no longer simply about storing ever-expanding quantities of data. The technology is increasingly expected to deliver broader strategic and operational outcomes. As AI makes cyberattacks faster and more scalable, this evolution becomes increasingly important. Enterprises must now invest in end-to-end resilience to protect the entire AI lifecycle, not just storage.

 

CIOs, CISOs and CTOs have every reason to be concerned. The challenge is not simply protecting enterprise data from attack. It is ensuring that when a cyberattack occurs, the enterprise storage can respond to its implications. In a very short amount of time, we have gone from a ‘when, and not if’ attitude towards the probability of a cyberattack occurring, to a ‘what will they do next?’ realisation. Now, thanks to AI, even cybercrime has been ‘democratised’ and brought into the realm of the everyday. That is why a cyber storage resilient, recovery-first strategy is becoming an increasingly important part of the enterprise response to an AI-accelerated cyber threat.

 

About Eric Herzog

Eric Herzog is the Chief Marketing Officer at Infinidat, a Lenovo company. Prior to joining Infinidat, Herzog was CMO and VP of Global Storage Channels at IBM Storage Solutions. His executive leadership experience also includes: Senior Vice President of Product Management and Product Marketing for EMC’s Enterprise & Mid-range Systems Division, and CMO and Senior VP of Alliances for all-flash storage provider Violin Memory.

 

www.infinidat.com

 

 

 

[1] https://www.boozallen.com/expertise/cybersecurity/threat-report-when-cyberattacks-happen-at-ai-speed.html