Two significant security issues surfaced this week involving very different parts of the Internet. Critical vulnerabilities were discovered in PHP, one of the most widely used technologies behind websites and web applications, while compromised cryptographic keys were discovered within eMudhra's Certificate Authority infrastructure. The technologies and circumstances are different, but both incidents are useful reminders of how much risk organizations inherit from technology they don't directly control.

On October 21, the PHP development team disclosed two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, that can allow unauthenticated remote code execution. PHP considers installations serving web requests to be at risk and specifically identified PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP and LiteSpeed. The project also stated that there are no complete mitigations and that organizations should update to a fixed version.

For most organizations, the immediate response will be to determine whether affected versions of PHP are running and get them updated. That is necessary, but it doesn't completely address the business problem because PHP may also exist underneath applications and services provided by someone else. An organization may be dependent on vulnerable technology without owning the system, managing the software or even having a clear view of where that dependency exists.

The eMudhra incident raises a different concern about the infrastructure used to establish trust on the Internet. On October 20, Google's Chrome Security team reported that private keys associated with eight eMudhra subordinate Certificate Authority certificates had been compromised. According to Google, insufficient cryptographic randomness during key generation made it possible to recover the private keys. Anyone possessing those keys could potentially create trusted HTTPS certificates to impersonate websites against systems that continued to trust the affected Certificate Authorities.

Mozilla's public incident record provides additional detail about the discovery. Mozilla confirmed the proof of key control, independently recovered the affected private keys and began blocking the compromised certificates in Firefox through OneCRL. Mozilla later determined that the circumstances justified emergency removal of trust in the emSign ECC Root CA - G3. eMudhra subsequently agreed to stop issuance from intermediates chaining to that root and said approximately 30,000 certificates would need to be migrated.

There is also important context around the actual impact. Google reported that it found no TLS certificates chaining to the six affected TLS intermediates in Certificate Transparency logs before the certificates were blocked. The discovery therefore shouldn't be interpreted as evidence that attackers were broadly impersonating websites across the Internet. The significance is that part of the infrastructure relied upon to establish trust had developed a weakness serious enough that browser vendors could no longer rely on that trust.

This is where I think these otherwise unrelated events become interesting from a risk perspective. We spend a lot of time determining whether security controls exist and whether they have been implemented correctly. Those questions still matter, but they don't necessarily tell us enough about what those controls depend on. An organization can correctly encrypt its communications while depending on a Certificate Authority to make that trust meaningful. It can patch every PHP system it manages and still rely on a provider running vulnerable software somewhere within a service the organization depends upon.

That dependency becomes increasingly important when we look at cyber risk from a business or insurance perspective. An application may show that encryption is enabled, vulnerability scanning is performed and patch management exists, but those answers only describe part of the risk. They don't necessarily tell us what happens when the technology or trust infrastructure underneath those controls fails outside the organization's environment.

What Should Individuals and Businesses Do Now?

For individuals, there isn't a complicated technical response required. The most important action is to make sure browsers, operating systems, phones and other Internet-connected devices are receiving their normal security updates. Google has already blocked the affected certificates for Chrome users through CRLSet and pushed equivalent protections to Android devices through Google Play Services. Google says most Android devices have already received the protection, while restarting the device can trigger an immediate Google Play Services update.

Businesses have a little more work to do. Organizations that operate websites, web applications or other Internet-facing systems should determine whether PHP is running anywhere in those environments and whether affected versions have been updated. The fixed releases are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. Organizations running older PHP versions should determine whether their operating-system or support provider has issued a backported security update or whether they need to move to a supported release. The PHP security advisory makes clear that updating is the only complete solution.

Businesses should also ask the same question of the companies managing technology on their behalf. A small or midsized business may not think it runs PHP because an outside company manages its website, e-commerce platform or web application. That doesn't remove the dependency. Asking hosting providers, web developers and managed service providers whether they are affected and whether they have patched vulnerable systems is a reasonable step right now.

For most business leaders, none of this requires becoming an expert in PHP or public key infrastructure. It does require knowing who is responsible for understanding these dependencies and whether they are being actively managed. A reasonable conversation with IT, your security team or your technology provider right now would be whether the organization is affected by either issue, whether vulnerable PHP systems have been patched, whether any certificates need to be replaced, and whether critical providers have been checked for the same exposure.

I don't see either of these incidents as evidence that the Internet's security model is falling apart. In fact, the response from PHP, browser vendors and others shows parts of that system working as they should. Problems were identified, fixes and protections were developed, and organizations now have information they can act on.

What I do think they expose is something worth paying attention to. Many of the controls we rely on have dependencies of their own. A certificate can validate correctly and encryption can be enabled while still depending on infrastructure outside the organization. A vulnerability-management program can be working well while the business remains dependent on software operated by someone else.

That matters when we evaluate risk because evidence that a control exists only tells us part of the story. We also need some understanding of what that control depends on and whether someone is paying attention when those dependencies change.

The practical response isn't panic. Patch the affected PHP systems, check the certificates that matter, make sure browsers and devices are current, and ask important technology providers whether they're affected. Then use incidents like these to get a little better at understanding what your business depends on. That's probably the more useful lesson.