17 Nov 2026Tue,

Rapid Healthcares
Infections

AI Revolutionizes Zero-Day Vulnerability Detection for Just $4

A recent study reveals that AI models can now identify zero-day vulnerabilities for as low as $4, impacting cybersecurity strategies.

The landscape of cybersecurity is evolving rapidly with advances in artificial intelligence (AI). Recent findings indicate that AI models are increasingly capable of identifying zero-day vulnerabilities for remarkably low costs, which raises significant concerns regarding cybersecurity defense mechanisms.

Affordable AI Detection

In a world where AI technology is advancing swiftly, the average price of utilizing these sophisticated models has dropped significantly. According to industry insights, token prices have faced a substantial decline, dropping an average of 47% per quarter since the beginning of 2023. Just last month, prices fell another 15%, driven by intense competition among leading AI developers such as OpenAI and Anthropic.

This decrease in cost is advantageous for businesses looking to harness AI-driven productivity and services. However, it has also made it easier for cybercriminals to execute attacks that exploit unknown vulnerabilities.

Study Overview: AI Performance in Vulnerability Detection

A report released by Xint.io provides an in-depth analysis of several AI models and their capabilities in detecting cybersecurity vulnerabilities. In the study, researchers investigated seven different AI models that were launched since February 2026, including two from OpenAI, three from Anthropic, and two open-source models from DeepSeek and Z.ai. They focused on how these models performed over time in identifying a specific vulnerability.

The vulnerability in question was a security flaw in Google’s V8 engine, designated as CVE-2026-102321. Discovered and disclosed by AI researcher Taisic Yun, this type confusion bug could lead to remote code execution (RCE).

Progression of AI Models in Detection Capabilities

The research commenced in February, initially spending a significant amount—$350—on testing to evaluate the best performing AI models. Shockingly, none of the models could uncover the crucial vulnerability at that time.

Fast forward to July, when GPT 5.6 Sol managed to identify the zero-day flaw in just 44 minutes at a cost of $22. By September, the more advanced GPT 6 Sol detected the same bug in a mere 13 minutes for only $4.24.

Out of the seven models reviewed, only three succeeded in locating the vulnerability: GPT 6 Sol, GPT 5.6 Sol, and Claude Opus 5. The latter found the flaw at a higher expense of $43.40 while requiring less effort compared to the others.

The Growing Threat of Inexpensive AI Tools

The research team identified a worrisome trend: as AI's capabilities surge, even less skilled hackers are able to exploit these advanced models to uncover vulnerabilities. This shift puts pressure on organizations to strengthen their cybersecurity measures, as what once felt out of reach is becoming accessible to mediocre hackers.

However, organizations can leverage the same AI technology to enhance their security posture. Security teams and penetration testers can run their own checks to safeguard systems before malicious actors can exploit weaknesses. An ongoing risk assessment focusing on weak points, such as public-facing services and unreviewed digital supply chains, is crucial to mitigate threats.

Conclusion: The Role of Human Expertise

Despite the impressive capabilities of AI in vulnerability detection, human expertise remains a vital component. Models often struggle to identify vulnerabilities without proper guidance, necessitating additional runs that increase both time and cost. As the researchers point out, while technical proficiency is required for software breach detection, the bar has been lowered significantly.

Nonetheless, it is essential for organizations to depend on not just AI technology but also to implement zero-trust access controls, bolster network defenses, and actively monitor endpoints to detect suspicious activities. By adopting a multi-faceted approach, companies can enhance their defenses against the looming threats posed by AI-assisted cyberattacks.

More from the newsroom

All stories