Analysis by cybersecurity firm Rapid7 highlights the challenges posed by AI-driven vulnerabilities, which are rapidly overwhelming traditional patch management strategies. Recent findings suggest that relying solely on patching is no longer sufficient to mitigate threats effectively.
Increased Vulnerability Disclosures
The second quarter of 2026 saw an unprecedented rise in vulnerability disclosures, with a significant increase in both volume and complexity. According to Rapid7’s report, titled ‘The Compression Era,’ the speed and precision at which attackers are exploiting these vulnerabilities are outpacing the ability of defenders to respond effectively.
Rapid7’s VP of cyber intelligence, Christiaan Beek, emphasizes that the driving force behind this surge is artificial intelligence. The report notes that high and critical vulnerabilities, rated between CVSS 7 and 10, doubled from 4,268 in Q2 2025 to 8,539 in Q2 2026. This rapid increase is attributed to AI’s capability to identify and exploit vulnerabilities swiftly.
AI and the Evolution of Cyber Threats
AI’s role in accelerating vulnerability discovery and exploitation is reshaping the cybersecurity landscape. New applications are constantly being developed, often containing previously identified vulnerabilities due to outdated coding practices. Beek points out that AI is perpetuating these issues by reusing old templates, further complicating the defense landscape.
This scenario is exacerbated by the asymmetry between attackers and defenders. Attackers need only find a single weak point, while defenders must secure a vast range of potential targets, from endpoints to APIs and supply chains. This widening gap between vulnerability disclosures and feasible triage capabilities underscores the urgent need for a shift in defensive strategies.
Nation-State and Criminal Cyber Activities
The report also sheds light on persistent cyber activities from nation-state actors such as China, Russia, Iran, and North Korea. These states target specific geopolitical adversaries, complicating the threat landscape further. Unlike financially motivated criminals, these actors often engage in prolonged espionage, leveraging sophisticated resources.
Ransomware remains a significant threat, with the United States being the most affected. In Q2 2026, there were 881 ransomware victims in the US, compared to 91 in Germany. Prominent ransomware groups include Qilin, The Gentlemen, and LockBit, with business services and healthcare being common targets.
The Need for Proactive Defense
Rapid7’s analysis illustrates that the rapid pace of AI-assisted attacks demands a proactive approach to cybersecurity. Beek suggests that reducing exposure rather than focusing solely on vulnerability severity can be more effective. Understanding which network areas are accessible to attackers and minimizing that exposure is crucial.
Traditional methods of assessing vulnerabilities through CVE scores are becoming obsolete. Instead, organizations need to focus on where vulnerabilities exist within their networks and the potential impact if exploited. By adopting this new perspective, defenders can better anticipate and mitigate threats.
The evolving nature of cyber threats necessitates a shift from reactive patching to proactive exposure management. As AI continues to transform the vulnerability landscape, organizations must adapt to maintain resilience in the face of increasingly sophisticated attacks.
