Skip to main content
AI finds zero-day flaws for just $4 now – even mediocre hackers can cash in - ZDNET

AI finds zero-day flaws for just $4 now – even mediocre hackers can cash in - ZDNET

Google News - DeFi/NFT/Web3
16 lis 2026, 19:17Google News
Oryginał
2 min read

The cost of discovering zero-day vulnerabilities has plummeted to as little as $4, making sophisticated hacking capabilities accessible to even mediocre hackers.

The cost of discovering zero-day vulnerabilities has plummeted to as little as $4, making sophisticated hacking capabilities accessible to even mediocre hackers. A recent report from Xint.io reveals how popular AI models have evolved to successfully find security flaws that would have been impossible to detect just months ago.

The research targeted a critical security flaw in Google's V8 engine (CVE-2026-102321), a type confusion bug leading to remote code execution that was not present in the AI models' training datasets. This vulnerability had been privately reported to Google's Vulnerability Reward Program by Xint researcher Taisic Yun.

The $4 breakthrough

When Xint first tested the vulnerability in February 2026, even the best AI models available couldn't find the flaw after 25 hours and $350 of scanning. By July, GPT 5.6 Sol discovered the zero-day flaw in just 44 minutes for $22. The most dramatic improvement came in September, when GPT 6 Sol found the same bug in just 13 minutes for only $4.24 with its first guided prompt.

While we are still far from complete amateurs using unharnessed models to hack systems and applications, what used to be considered too hard and costly to be worth the effort of the average hacker is now reachable.

Xint researchers

Model performance varies

Of the seven AI models tested—two from OpenAI, three from Anthropic, and two open-source models from DeepSeek and Z.ai—only three successfully found the vulnerability: GPT 6 Sol, GPT 5.6 Sol, and Claude Opus 5, all launched from July onward. The older models and open-source options failed the task entirely.

GPT 5.6 Sol required two prompts to find the bug, while Claude Opus 5 succeeded on its first try but at a higher cost of $43.40. The researchers noted that expertise still matters when instructing AI models to discover vulnerabilities, as some models couldn't find the target without additional guidance, increasing both time and cost.

Security implications

The falling cost of AI tokens—down by about 47% per quarter since 2023 and 15% last month alone due to competition among major AI developers—has made these powerful tools more accessible to both attackers and defenders. The researchers found that software breaches still require some level of technical proficiency, but no longer top-of-the-distribution expertise.

Organizations now face the challenge of adjusting their security postures to this new reality. However, the same AI tools can be used defensively to find vulnerabilities before threat actors do. By focusing risk assessments on likely weak spots such as public-facing services or unaudited digital supply chains, companies can mitigate the threat of AI-enabled cyberattacks.

Z.ai
Article written by AI
Z.ai GLM-4.5-FlashFree
Free AI articles150 of 150
No limit after signing in with Telegram