'Every single week it's going to feel like the sky is falling': Cloudflare CEO says strap in for 'a couple years' of AI-powered internet security mayhem

Every week, it feels increasingly like the internet's very foundation is crumbling down, according to Cloudflare's CEO Matthew Prince. During a Reuters interview, Prince warned that in the coming years, we can anticipate "a couple years of AI-powered internet security mayhem" as large language models (LLMs) accelerate the discovery of exploitable security vulnerabilities.

Cloudflare, a leading cloud networking and cybersecurity provider that powers countless websites, recently discovered a handful of severe vulnerabilities in PHP—a programming language utilized by around 70% of the world's websites. These vulnerabilities put popular web destinations at risk of being weaponized by malicious actors.

The rapid identification and exploitation of these security flaws have been largely hampered by the limited number of cybersecurity experts and the time and expertise required to develop working exploits. However, with the advent of powerful LLMs, the time and resources needed for this process have drastically reduced. Wiz Research, a company that utilized LLM tools to identify the PHP vulnerabilities, noted that these frontier models made the process practical, enabling them to develop and adapt working exploits across multiple applications within hours.

Prince emphasized that while the short-term consequences will undoubtedly be challenging, the long-term outlook remains positive. As these vulnerabilities are patched and the security of software improves, we can expect the internet to become more robust against future attacks. For now, Prince advises users to stay vigilant with regular browser and app updates and maintain an appropriately heightened sense of caution.