Saved To My Saved Content
Use BCG Answer to summarize this article

The alarm bells started ringing on Oct. 21, 2026, with the disclosure of critical vulnerabilities in PHP, a programming language used by roughly 70% of websites worldwide. Attackers had already begun attempting to exploit the flaws days before fixes became available.

Reports of other cyber incidents came in quick succession, including malicious updates to an AI software package, compromised digital certificates, and hackers using AI agents to gain unauthorized access to corporate systems.

The “Black October” wave of cyber incidents has continued into November, heightening concerns about the growing power of AI-enabled attacks. But it has also exposed a deeper problem: the enormous technical debt companies have accumulated over the years.

Many organizations have been living with outdated software, aging infrastructure, and systems that are difficult or even impossible to update. Now AI is making it much easier for attackers to exploit these weaknesses, and vulnerabilities that companies once considered manageable have become a critical business risk.

How is AI driving the recent increase in cyber attacks?

AI is reducing the time, cost, and expertise required to execute sophisticated cyber attacks. Tasks that once demanded considerable technical skill can increasingly be automated, allowing attackers to pursue more targets simultaneously.

Cybersecurity and IT leaders echoed this in BCG and GLG's Black October Survey 2026, fielded in early November: nine in ten respondents said AI is changing how software vulnerabilities are discovered and exploited.

“It’s actually changing the economics of a hack,” says Vanessa Lyon, a BCG managing director and senior partner and the founder and director of BCG’s Center for Leadership in Cyber Strategy. “A major hack used to take several hours or several days, perhaps two weeks for something more complicated. And it often wasn’t worth pursuing because it required too many resources. Now attackers can move much faster without getting caught, or carry out many more attacks in parallel.”

A major hack used to take several hours or several days, perhaps two weeks for something more complicated. Now attackers can move much faster without getting caught. — Vanessa Lyon, Managing Director & Senior Partner

AI also enables more elaborate attacks. Criminals can use information stolen from an organization to impersonate employees and create convincing deepfakes. The growing availability of powerful AI models compounds the challenge: generally accessible open-weight models are becoming increasingly capable, putting sophisticated hacking tools within reach of a wider range of attackers.

For companies, the result is a shrinking window to detect and respond to threats. “Attacks can be basically over before the attacked organization knows what happened,” says Stefan Deutscher, a BCG partner and director and the firm’s global topic leader for cybersecurity and IT risk management. “In this environment, defending at machine speed becomes paramount.”

Subscribe
Stay ahead with BCG insights on artificial intelligence

Why is technical debt a cybersecurity risk, and how does AI compound the risk?

Technical debt, the backlog of outdated software and infrastructure that organizations put off fixing, leaves known weaknesses that attackers can exploit—and AI helps them find these weaknesses faster.

Replacing aging applications and infrastructure is expensive, disruptive, and often difficult to justify when existing systems continue to function. Many organizations operate software that hasn’t been updated for years, including custom-built applications that few employees understand. “It still runs, but nobody knows why it runs, and nobody wants to touch it,” Deutscher says.

Some legacy systems are so outdated that they cannot be readily patched. The problem is especially acute in manufacturing and other industries that depend on operational technology, where replacing aging software or hardware can require shutting down production.

Previously, exploiting vulnerabilities in these systems required extensive research and specialized knowledge. AI makes it easier for attackers to analyze unfamiliar code and identify weaknesses.

Complex technology environments make it harder to coordinate remediation across different parts of the business.

Complex technology environments also make it harder to coordinate remediation across different parts of the business. Security teams may identify a vulnerability but lack the authority to address it without coordinating with IT and operations. Competing priorities and lengthy approval processes also can create challenges.

This can result in slower response times in the event of an attack. In BCG and GLG’s survey of cybersecurity and IT leaders, about 90% of respondents from companies with fewer than 500 employees reported addressing their most critical issue within three days, compared with roughly 60% at companies with 10,000 or more employees.

The difference may have less to do with a company’s size than with the complexity of its technology environment and decision-making processes. Even so, larger organizations may be able to learn from the agility of their smaller counterparts, particularly when it comes to streamlining decisions and coordinating remediation in the face of an attack.

What other cybersecurity risks do AI agents create for businesses?

AI is not just making it easier for external attackers to exploit existing vulnerabilities. As companies deploy agentic solutions throughout their operations, they are also introducing new risks from within.

AI agents can access sensitive information, communicate with other applications, and take actions without continuous human supervision. These agents are “very capable, very fast, and typically not yet well regulated and governed,” Deutscher warns. And they may take actions their developers did not anticipate.

Consider an AI agent tasked with analyzing a company’s competitive position. In seeking information from another company’s agent, it might inadvertently disclose confidential pricing data. An attacker could also compromise a third-party agent and exploit its access to internal systems.

Companies need to understand which agents operate within their environments, what information they can access, and who is responsible for controlling their behavior. Without appropriate safeguards, they risk introducing new weaknesses even as they work to eliminate old ones.

Will a bigger cybersecurity budget solve the problem?

Not on its own. In the BCG and GLG survey, 69% of respondents expected cybersecurity budgets to rise above their existing plans following the October incidents. But Lyon argues that companies need to reconsider where that money goes. “The problem is not the cyber budget, the problem is actually the IT budget,” she says.

Cybersecurity tools remain essential for detecting threats, protecting systems, and responding to incidents. But they cannot eliminate the underlying vulnerabilities created by outdated infrastructure and poorly maintained applications.

Lyon compares the problem to the story of the Three Little Pigs. “If you have a house made of straw, regardless of what you do, the wolf will come and blow it down.”

Companies must strengthen the foundations of their technology rather than simply relying on additional layers of protection. That means allocating more of their broader IT budgets toward modernizing legacy systems, remediating vulnerabilities, and improving the quality of new software. These decisions require coordination across the C-suite: CIOs and CTOs must work with CEOs, CFOs, and other leaders to determine which modernization investments are most necessary.

As they make these critical updates, companies can avoid creating additional technical debt by factoring in maintenance and technical refresh rather than implicitly making it someone else’s problem a few years down the road. After all, nobody would build a house without also budgeting for occasional repairs and renovations.

How can business leaders make their companies more cyber resilient?

Modernization takes time, and no organization can eliminate every vulnerability. Companies must reduce their exposure while strengthening their ability to withstand attacks that succeed. Leaders should prioritize four actions:

AI will continue to make cyber attacks faster and more accessible. The companies best positioned to withstand them will be those that confront existing weaknesses in their technology, limit new ones, and make hard choices about where to invest—and prepare for the moment when even the best cyber defenses will fail.