A web technology that has been around for decades came close to causing a global incident. PHP, the open-source programming language behi...
A web technology that has been around for decades came close to causing a global incident. PHP, the open-source programming language behind a huge share of websites, had two security flaws that sat unnoticed for years. Researchers recently worked out how to exploit them, and the internet had to scramble.
The flaws are called PHP2Shell and are tracked as CVE-2026-109618 and CVE-2026-109619. In plain terms, anyone who knows how to exploit them can break into an online system running PHP without a password. That covers websites, internet-facing servers, and anything else built on the language. Once inside, an attacker can read or change the data stored there.
It was a big event. Millions of people around the world rushed to patch their systems before malicious hackers could get to them first.
What happened and how the bugs were found
Security researchers at Calif found the two bugs in PHP, a language used to build websites and web apps. It runs on the public internet and also inside company networks, on intranets and internal sites.
PHP is open source, which means anyone can use it for almost any purpose at no cost. That is a big reason it became one of the most widely used web programming languages.
The bugs were first spotted in 2024, but nobody thought they could be exploited at the time, so they were never patched. That changed when researchers recently found a way to use them. Calif then told the PHP Foundation, the group that maintains PHP.
The foundation started fixing the flaws under a coordinated disclosure process. It shared details and patches under a tight embargo, meaning the information was kept confidential until a set deadline. This gave some of the largest internet companies and web hosts that depend on PHP a head start on installing the fixes. On Wednesday, the foundation released the patches to everyone.
The weeks of quiet coordination paid off. So far, the world has mostly avoided mass exploitation. Governments began urging citizens and businesses with websites to take action. Banks, retailers, gaming sites, online stores, hospitals, and some government agencies pulled servers offline to patch them.
Security researchers still warn that the situation is not over. A large number of systems around the world remain unpatched and open to hacking and data theft.
Why the PHP2Shell vulnerabilities are so serious
Several things make these bugs dangerous.
They can be abused remotely over the internet. The victim does not have to click a link or open a file. Security professionals call this remote code execution, and it ranks among the most damaging kinds of bug because it can give an attacker, anywhere in the world, effectively full control of an unpatched system.
The bugs also allow unauthenticated access. An attacker does not need a password or any other credential to get in. Any unpatched system reachable from the internet is a potential target.
After breaking in, an attacker could plant malware to steal data, or simply leave a hidden foothold to return to later. Either outcome is bad for a business, a hospital, or a government agency.
Why 70% of the web was exposed
PHP is wildly popular. It is free, relatively easy to learn, and has been around for decades. Roughly 70% of the web-facing internet relies on it, although its share has been shrinking slowly as developers move to newer and more secure languages.
Many of the best-known website platforms run on PHP, including WordPress, Drupal, Joomla, and MediaWiki, the software behind Wikipedia. A flaw in PHP therefore affects every site built on those platforms.
PHP is one of many aging but functional technologies that have outlasted several waves of the internet. They are so common now that they act like glue holding much of the web together. Open-source maintainers often build these tools without pay, so they cannot afford third-party security reviews. Many companies rely on them heavily and never contribute money to the projects.
Was AI used to find the bugs?
Calif was credited with discovering the bugs earlier this year during a security audit of PHP's code. The team was taking part in Hack the Planet, an initiative from AI model maker OpenAI and cloud security company Wiz. Wiz later verified the proof-of-concept exploits Calif built.
Calif's own write-up, which includes comics and is worth reading, says AI helped with exploiting the bug. Most of the work, however, came from what the researchers already knew. They understood PHP's codebase and its history of exploitation well. The team also made clear that no AI model can simply find every bug from a few prompts, at least for now.
Were the bugs exploited before disclosure?
The embargo was meant to keep the details quiet until Wednesday, but some exploitation appears to have started earlier.
Amazon, which sees a large amount of internet threat activity because of its scale, said it detected PHP2Shell exploitation about a week before the patches came out. That suggests someone knew how to exploit the bugs before the details were public. Amazon could not say how the attacker learned about them, and it advised defenders to look for signs of exploitation going back to October 17.
There are a few theories about what happened:
Patch details leak the bug when a fix is published for open-source code, anyone can inspect it. Exploiting the bug may be hard, but seeing where it sat and how it was fixed shows attackers where to aim.There may have been a leak or a communication breakdown during the embargo. The open-source community is enormous, and keeping information contained depends on good faith.
Attackers may have found the bugs on their own, or obtained them through espionage.
The PHP Foundation blamed one company for pushing the patches live hours before the embargo ended. That could have given some attackers a head start if they got hold of the patch early. It does not explain the week of activity Amazon saw, though.
Researchers say turning these bugs into working attacks takes a lot of effort. That is difficult but not impossible, and it helps explain why mass exploitation has not appeared yet. Even so, as of late Friday, security companies Palo Alto Networks and Trail of Bits confirmed real compromises and post-exploit activity tied to the PHP bug. Attackers have already used it to break into at least one target and carry out malicious activity there. The identity of that target is unknown.
Did the internet dodge the worst of it?
Large parts of the internet went dark during the patching rush and are now coming back. The absence of mass exploitation so far is an encouraging sign, but the danger has not passed.
Matthew Prince, chief executive of network security firm Cloud-flare, said this week that Cloud-flare has seen an exponential rise in attempts to exploit the bugs since they became public. Cloud-flare has visibility into a large portion of internet traffic, so the figure shows how fast malicious actors moved.
PHP also does not live only on web servers. It appears in developer tools, command line and terminal packages, and software components used by routers, storage devices, and enterprise security equipment such as firewalls and VPNs.
Exploitation will probably continue for some time. Smaller IT teams and solo developers tend to update slowly. Companies that build PHP into their own products will also need to ship updates before their customers are protected.
As more systems get patched, the attack surface will shrink. Still, the effects could last for years, especially in legacy systems that have no way to be updated.
Will something like this happen again?
Internet-stopping days are rare.A security reporter, recalls only a few over the years. In 2016, a massive botnet flooded the internet registry Dyn with junk traffic and slowed many of the web's top services to a crawl. In 2017, North Korea used leaked U.S. government hacking tools to infect the U.K.'s health system with ransomware. In 2024, a faulty CrowdStrike update caused a global outage that crashed most of the world's Windows computers at once.
PHP2Shell could have been another day like those. It probably will not be the last major incident, and it may mark the start of a trend of attacks on legacy technology. Two forces drive that trend. Aging, widely used software still hides flaws nobody has found. And AI tools are making it easier to find them.
Memory-unsafe languages and the shift to safer code
PHP, like other web technologies from decades ago, is memory unsafe. That means it is open to a whole class of vulnerabilities that can force software to leak sensitive data from its memory. The National Security Agency says around 75% of actively exploited vulnerabilities abuse memory-related flaws, so hackers use them often.
This is one reason developers are moving to memory-safe languages such as Python, Rust, and Swift. These languages cannot be abused in the same way, which makes them largely immune to a wide range of vulnerabilities that still affect older languages.
A large part of the internet still runs on technologies from decades ago, though. PHP is one. Media and video codecs and compression libraries are others. When any of them breaks, the effects reach much of the rest of the web.
What website owners should do now
Update PHP to the patched release as soon as possible, including on any product or device that bundles PHP.Check logs for signs of exploitation going back to October 17.Update platforms built on PHP, such as WordPress, Drupal, Joomla, and MediaWiki, as their maintainers release fixes.Ask vendors of routers, storage devices, firewalls, and VPNs whether their products include PHP and when updates will arrive.Plan to replace or isolate legacy systems that cannot be patched.
Frequently asked questions
What is PHP2Shell?
PHP2Shell is the name given to two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619. They let an attacker break into a system running PHP without a password and gain control of it remotely.
Who discovered the PHP2Shell vulnerabilities?
Security researchers at Calif found them during a security audit of PHP's code. Wiz verified the proof-of-concept exploits.
Are the PHP2Shell bugs fixed?
Yes. The PHP Foundation released patches on Wednesday. Systems only stay protected if administrators install them, and many have not yet.
Why is remote code execution so dangerous?
It lets an attacker run commands on a system from anywhere in the world, without the victim clicking anything. With full control, an attacker can steal data or install malware.
Do I need to worry if I use WordPress?
WordPress, Drupal, Joomla, and MediaWiki all rely on PHP, so you should apply updates for both PHP and your platform as they become available.
Was the vulnerability exploited before the patch?
Amazon reported seeing exploitation about a week before the patches were released and told defenders to check for activity from October 17. Palo Alto Networks and Trail of Bits have since confirmed real compromises.
Did AI find the bugs?
No. AI helped with the exploitation work, but the researchers say most of it relied on their own deep knowledge of PHP.
Can PHP2Shell affect devices other than websites?
Yes. PHP is also used in developer tools and in components of routers, storage devices, and enterprise security equipment such as firewalls and VPNs.
Will there be more attacks like this?
Probably. Aging, memory-unsafe software likely still hides undiscovered flaws, and AI tools are making them easier to find.
References
Tech crunch - https://techcrunch.com/2026/10/24/php-ocalypse-how-two-critical-web-bugs-almost-broke-the-internet/
