Security teams and web administrators rushed this week to patch newly disclosed PHP flaws, while criminal hackers moved quickly to exploit them in the wild. Cloudflare’s analysis shows millions of exploitation attempts within hours of the vulnerabilities becoming public, a result that is unsurprising given the widespread availability of AI tools that can accelerate attack development.
So far, the feared scenario of massive victim takeovers has not materialized. The PHP Foundation reports no confirmed successful exploits of the bugs, collectively labeled PHP2Shell, and many of the publicly shared exploit code samples appear ineffective. Nonetheless, the most notable aspect of this week’s turmoil is how advanced AI models assisted researchers in turning a long‑standing vulnerability into an immediate crisis.
Researchers at the security firm Calif were the first to spot the PHP2Shell issues and craft exploits in early August. They used cutting‑edge AI tools as part of OpenAI’s Patch the Planet program, combined with deep human expertise in PHP’s architecture and weaknesses. Some of the underlying problems had been identified in 2024 but were considered too complex for exploitation at that time. After Calif’s findings, teams at OpenAI and the Google‑owned security company Wiz independently uncovered the same flaws using AI assistance.
“The vulnerability had been known to the PHP team for years, but because it could not be leveraged by an attacker – at least not a human one – it was not addressed,” says Sagi Tzadik, a security researcher at Wiz. No software is flawless, and even the most fundamental digital tools contain defects. The growing power of AI models has shortened the time needed to analyze and potentially exploit those defects, increasing pressure on under‑resourced IT staff, developers, and administrators. Microsoft’s recent security bulletin listed a record 1,453 vulnerabilities for the month, and Bank of America hired 1,000 consultants to remediate issues uncovered by Anthropic’s Mythos.
Security experts stress that AI is not inventing new threats; it is amplifying existing patterns of vulnerability discovery and coordinated disclosure. “We’ll definitely see more of these long‑life bugs that are in every version of the software for the last decade or whatever,” says Jake Williams, vice president of research and development at Hunter Strategy and a former NSA hacker. “But I think there’s only so much of this low‑hanging fruit. There isn’t much else that has the specific reach of PHP. It’s a pretty wild attack surface – most other tools aren’t exposed to the internet by design.”
Researchers also note that even with frontier AI models, sophisticated hacking still relies heavily on human skill. Ryan Austin, the Calif researcher who initially identified the PHP flaws, wrote this week, “No model is simply capable of unearthing ‘all bugs’ with simple prompts. At least for now.” In short, security engineer and researcher Niels Provos suggests the panic can be tempered. “To be honest, I am not really worried,” he says about AI uncovering a flood of historic bugs that become exploitable. “People have now been looking with AI for over a year, and it required a highly skilled team to find this one. I guess we will see whether I am completely wrong, but I doubt it.”