26.10.2026, Taganrog.
The capabilities of generative artificial intelligence (AI) to find vulnerabilities in software have already become a game changer in cybersecurity, and not in favor of security, Rossa Primavera News Agency‘s IT Desk notes.
On October 23, OpenAI informed that in September its AI agents, which during testing were forced to look for ways to break out of an isolated environment, along the way found a critical hole in the popular software component Brotli, which is used by many websites and servers. And this is not an isolated case.
A few days earlier, on October 20, it became known that researchers using AI had found two critical holes in PHP, the language on which most websites run. The vulnerabilities allowed servers to be seized.
To prevent hackers from exploiting the holes in PHP before patches were installed, the governments of many countries disabled a number of state internet platforms. Commenting on this news, the deputy director for information security at Anthropic warned of an impending cybersecurity crisis.
“The vulnerability tsunami is coming in just a few months and there won’t be enough time to patch critical infrastructure,” he wrote on the social network X.
Thus “traditional” cyber defense is physically unable to keep up with automated searching for security holes. This does not mean that one should give up. It means that non-trivial protection measures must be devised, otherwise the “tsunami” may sweep away those who continue to work at the same pace.
Source: Rossa Primavera News Agency
