Cutting both ways: AI sharpens systems protection while opening new exploits for malicious actors

The cybersecurity landscape is undergoing a transformation as powerful frontier AI models are being embraced for their ability to identify vulnerabilities with unprecedented speed. However, this dual-edged sword also presents new dangers, as malicious actors can leverage AI models to uncover and exploit weaknesses. A recent example of this duality can be seen in the PHP programming language, which powers roughly 70% of global websites.

Researchers utilized OpenAI's GPT-5.6 Sol to scan PHP's code base and identified critical vulnerabilities, prompting organizations to scramble for patches. At the same time, Amazon disclosed that attackers began targeting PHP vulnerabilities just four days before patches were released. Cloudflare data has since seen a surge in exploit attempts, reaching into the millions.

This ongoing digital threat landscape underscores the necessity for organizations to be prepared for inevitable breaches. During Dell Technologies World in Las Vegas, attendees were reminded that data protection and cyber resilience have become foundational control layers for enterprise AI. Krista Case of theCUBE Research highlights the need for organizations to adapt security controls to the AI era, encompassing cyber recovery capabilities, AI governance, data integrity, and operational trust.

However, there is a growing concern within the cybersecurity community that AI may pose a significant threat to systems protection. Microsoft's monthly report revealed 1,453 new vulnerabilities in October alone. Steve Kenniston, senior cybersecurity evangelist for Dell, explains that AI changes the game with model inferencing, training data, prompt injection, and identity management considerations.

Dell is actively employing frontier models to build defenses, participating in Anthropic PBC's Project Glasswing to secure critical systems. This initiative aims to bring together organizations whose software supports critical infrastructure, protecting against catastrophic attacks.

The rise of AI agents has also increased scrutiny on identity in enterprise systems, as easier access to AI tools enables threat actors to bypass traditional access management and detection systems. John Scimone, president and chief security officer of Dell, notes that the cost of entry for attacks has dropped dramatically, making them more sophisticated than ever before.

Another critical aspect highlighted by InfoScale's Joseph D’Angelo is the lack of warning in the face of sophisticated attacks. Unlike natural disasters, major cyber attacks often strike without any indication of impending danger. This necessitates a defensible recovery posture, focusing on understanding the state of applications, data, and being able to recover without compromising the attack.

Data plays a pivotal role, as AI systems powering enterprise applications are only as good as the data that fuels them. Companies like Infinidat are developing security solutions to protect operational data as AI increases the speed and complexity of cyber risks, including immutable snapshots, cyber detection, and recovery capabilities.