Major Security Breaches and the Role of Artificial Intelligence

October 2026 witnessed a series of major cybersecurity incidents, prompting experts to describe the period as “Black October”. The incidents exposed vulnerabilities in digital infrastructure and highlighted how artificial intelligence (AI) is transforming both the discovery and exploitation of software flaws.

On October 17, researchers reported weaknesses in private cryptographic keys used in eMudhra certificates, potentially allowing attackers to impersonate legitimate websites. On October 21, the PHP Foundation announced two critical vulnerabilities, known as PHP2Shell, that could enable remote attackers to compromise vulnerable PHP servers. Around the same time, the open-source AI development repository LiteLLM was compromised, while GitLab issued fixes for a highly severe security flaw.

AI-assisted research played a significant role in discovering PHP2Shell. Researchers associated with AI firm Calif and cybersecurity company Wiz used advanced AI models to identify vulnerabilities that had remained difficult to reproduce for years. PHP is widely used to develop websites and web applications, making these vulnerabilities potentially far-reaching.

However, details of the exploit reportedly became public before a patch was available, increasing the risk of attacks against vulnerable systems. The incident also exposed the challenges faced by open-source maintainers, many of whom depend on limited resources and volunteer contributions to develop and distribute security fixes. Even after patches were released, older PHP versions remained a concern.

The episode has renewed debate over responsible vulnerability disclosure, which seeks to balance the need to inform developers with the risk of giving attackers an opportunity to exploit security weaknesses before systems are protected.

AI’s Promise, Limitations and Ethical Concerns

AI is increasingly being used to identify software vulnerabilities, analyse code and improve cybersecurity testing. Anthropic’s Cyber Verification Program, for instance, reported using formal verification techniques to identify bugs in open-source software libraries that had escaped detection despite extensive testing.

Formal verification uses mathematical methods to assess whether a system satisfies specified requirements. AI could help make these techniques more accessible and efficient. However, the process has limitations: the requirements must be correctly translated into mathematical models, and the results depend on the assumptions underlying those models. A successful proof therefore does not guarantee that software is free from every possible security flaw.

The growing role of AI also raises ethical and economic concerns. Companies developing advanced AI systems could potentially sell tools that identify vulnerabilities while also offering products designed to defend against them. This creates questions about conflicts of interest, accountability and whether commercial incentives could influence cybersecurity priorities.

The expansion of AI infrastructure raises additional environmental concerns because data centres require substantial energy and computing resources. Meanwhile, technical claims about AI’s capabilities require independent verification and careful human assessment.

The recent incidents demonstrate that AI can strengthen cybersecurity but can also accelerate the discovery and exploitation of vulnerabilities. A safer digital ecosystem will require timely security updates, responsible disclosure, adequately supported software maintainers, independent testing and continued human oversight.