# AI Is Accelerating Cyber Threats. Do Enterprises Know Where They're Exposed?

> October's security reports exposed a growing challenge for enterprise technology leaders: finding a vulnerability is one thing. Understanding its impact across an entire software estate is another. 

- Published: 5 Nov 2026
- Category: Perspectives
- Author: CohesionX
- Reading time: 2 min

_October's security reports exposed a growing challenge for enterprise technology leaders: finding a vulnerability is one thing. Understanding its impact across an entire software estate is another._

Artificial intelligence is changing the pace of cybersecurity. Attackers are increasingly using AI to automate activities that previously required considerable time and specialist effort.

In September 2026, Google's Threat Intelligence Group [reported](https://cloud.google.com/blog/topics/threat-intelligence/from-prompting-to-autonomy-the-evolution-of-adversarial-ai?) that adversaries were progressing from basic AI prompting to agentic workflows and automated cyber operations. In one observed campaign, attackers used AI-enabled processes to execute a large-scale credential-harvesting operation in under six hours.

Meanwhile, October delivered another wave of serious security advisories involving widely used technologies.

## **Five warnings from October 2026**

The following are just five examples flagged during the month, **not an exhaustive account of October's vulnerabilities or cyber incidents.**

![Table summarizing recent security issues and their potential impacts. Listed issues include PHP Core vulnerabilities (CVE-2026-109618/109619) enabling remote code execution, a GitLab vulnerability (CVE-2026-110264) that may allow unauthenticated code execution, a Node.js/Brotli vulnerability (CVE-2026-111497) causing memory corruption, information disclosure, or remote code execution, an Unbound DNS-over-QUIC vulnerability (CVE-2026-97206) leading to denial of service, and weak cryptographic keys affecting eMudhra certificate authorities that could enable certificate impersonation.](https://cdn.sanity.io/images/9z26qkwl/production/fa048a9c8049a60b7fdbbea14dd457840dd27be5-877x276.png)

These issues differ in their causes, exploitation requirements and consequences. They were not all AI-generated or AI-driven.

What connects them is the challenge they present to organisations operating complex technology environments.

## **The vulnerability is only the beginning**

Consider an enterprise running hundreds of applications, databases, APIs and integrations developed over many years.

A critical vulnerability is announced. Security teams receive an alert.

**Now what?**

Which systems contain the affected component? Which applications depend on it? Could a compromised service expose customer information, interrupt operations, or affect other business-critical systems?

For many organisations, answering these questions means searching through fragmented documentation, consulting developers and manually tracing dependencies.

A time-consuming exercise.

And time is increasingly important when attackers can use automation to accelerate parts of the attack process.

## **Defensive AI needs to keep pace**

This is where AI can play an equally important role in defence.

Products such as [**Runestone**,](https://cohesionx.co.za/products/runestone) developed by South African enterprise AI company CohesionX, address a fundamental problem: enterprises often lack a complete, current understanding of the software they already own.

Runestone uses read-only analysis to map source code, databases, APIs, configurations and integrations into a connected, continuously updated view of the software estate.

When an issue is identified, its dependency mapping and blast-radius analysis help teams understand which systems could be affected. Findings can be supported by vulnerability intelligence, historical logs and other evidence, then routed to the appropriate owners for investigation.

It also maintains living documentation and audit trails, helping organisations retain critical technical knowledge and strengthen governance.

Runestone does not replace patching, security controls or human judgement. It provides the context needed to make those activities more effective.

## **You cannot secure what you do not understand**

AI is accelerating software development and parts of the cyber threat landscape. Enterprise security strategies need to evolve accordingly.

CIOs and CISOs can no longer simply ask whether their organisations can detect vulnerabilities. **It is whether they understand their exposure well enough to respond decisively.**

[Discover Runestone: Enterprise software intelligence and defensive AI](https://cohesionx.co.za/products/runestone)