Key Takeaways
- The PHP security team released fixes for two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, on 21 October 2026 at 13:37 UTC (09:37 EDT). Both carry a CVSS v4.0 score of 9.2 and can lead to unauthenticated remote code execution.
- The vulnerabilities involve memory corruption during out-of-memory handling and a use-after-free condition caused by stale global state between requests. Web-facing PHP deployments, including Apache mod_php and PHP-FPM, are affected, although exploitation requirements differ by configuration.
- Public exploit code appeared before the official release, with reported successful remote code execution and ASLR bypass against Apache mod_php. This increases the risk of attacks against unpatched systems. Amazon recorded an exploitation attempt on 17 October, four days before the fixes, and Cloudflare reported thousands of attempts in the hours before release. These are attempt counts, not confirmed compromises.
- Fixes are available in PHP 8.2.35, 8.3.36, 8.4.27 and 8.5.12. PHP 7.4, 8.0 and 8.1 are also affected and have no official release, and older versions have no fix. Organizations using vendor-maintained PHP packages should check for backported security fixes.
- No complete workaround is available. Organizations should prioritize patching internet-facing PHP deployments immediately, verify that updated runtimes are actually in use, and investigate potentially compromised systems. Temporary configuration changes may reduce exposure but do not replace patching.
TrendAI™ customers can visit this knowledge base entry for information on available solutions and how to mitigate these vulnerabilities.
What happened
On 20 October at 15:20 UTC, a Hacker News post leaked details of two critical PHP vulnerabilities that could allow remote code execution. The post was removed, but screenshots circulated on X and Mastodon. Security researchers at watchTowr subsequently confirmed the vulnerabilities and identified them as CVE-2026-109618 and CVE-2026-109619. Coordinated disclosure had originally been planned for 22 October.
Later, on 20 October, the PHP security team published an advance notice warning of the vulnerabilities, without disclosing technical details or affected versions.
On 21 October at 13:37 UTC, PHP released security updates for both vulnerabilities, each rated 9.2 under CVSS v4.0. Both can lead to unauthenticated remote code execution across multiple PHP deployment configurations. Fixes are available for all four supported PHP branches, from 8.2 through 8.5.
Public exploit code also appeared before the official release, demonstrating a potential path to remote code execution against vulnerable Apache mod_php deployments. Organizations should now prioritize patching and investigating potentially compromised systems.
On 23 October, PHP published an informational advisory for a separate vulnerability, CVE-2026-115560, in the same session upload progress feature. PHP describes it as an unlikely remote code execution vector that requires an unusual configuration, and it will be fixed in the next regular PHP release.
On 29 October, Dream Research reported that a suspected China-aligned actor chained these PHP flaws with a Linux kernel privilege-escalation vulnerability, Fragnesia ( CVE-2026-46300 ), to gain root on 25 Palo Alto Networks PAN-OS firewall and management servers in the US, Japan, Taiwan and South Korea. The servers were exploited on 23 and 24 October, and Dream says the full chain, from a single unauthenticated HTTPS request to a persistent root webshell, takes under 15 minutes. The affected servers included three in a US government-affiliated environment. In its 29 October update , Palo Alto Networks’ Unit 42 independently verified the attacker’s tooling and Dream’s reported results. Palo Alto Networks has released fixed PAN-OS versions and advises removing internet access to management interfaces .
Vulnerability details and current status
| Release time | 21 October 2026, 13:37 UTC (09:37 EDT), confirmed by PHP. |
| Severity | Critical. CVSS v4.0 score of 9.2 for both vulnerabilities, confirmed by PHP. |
| Type of flaw | Both vulnerabilities can lead to unauthenticated remote code execution. CVE-2026-109618 involves memory corruption during out-of-memory handling; CVE-2026-109619 involves a use-after-free condition caused by stale global state persisting between requests. |
| Affected versions | PHP 8.2 releases before 8.2.35, 8.3 before 8.3.36, 8.4 before 8.4.27, and 8.5 before 8.5.12. PHP 7.4, 8.0 and 8.1 are also affected and receive no official release, and PHP 7.3 and older, including PHP 5, are affected with no fix available. PHP has provided source-level backports for 7.4, 8.0 and 8.1; vendor-maintained packages may receive separate fixes. |
| Configurations affected | PHP confirms that the vulnerabilities affect web-facing deployments, including Apache mod_php, PHP-FPM, CGI, FrankenPHP and LiteSpeed. Demonstrated exploitation paths and prerequisites vary by vulnerability and configuration. Apache mod_php and PHP-FPM have documented remote code execution paths. CLI deployments are not directly exposed to HTTP-based exploitation but may be affected when processing attacker-controlled input. |
| Exploitation in the wild | Amazon Threat Intelligence recorded an exploitation attempt against its honeypots on 17 October, before the fixes were released, and more on 21 October. Cloudflare reported thousands of attempts in the hours before release, rising to tens of thousands within 45 minutes of the announcement. watchTowr also reports exploitation attempts in its honeypot network. As of 25 October, TrendAI™ has not observed a confirmed successful compromise or an attributed threat actor campaign. On 26 October, both vulnerabilities have been added to CISA’s Known Exploited Vulnerabilities (KEV) catalog with a 29 October remediation deadline for federal agencies. |
| Public proof of concept | Multiple public exploits are now available. Rapid7 found that each works only against a specific combination of operating system, PHP version and web server, and none appears broadly reliable. Sansec reports that the public exploits do not work against a default PHP installation; most require a separate file inclusion flaw in the target application. |
| Workaround | No complete workaround exists. PHP recommends temporary risk-reduction measures, including enforcing memory_limit=-1 with resource controls for CVE-2026-109618, and disabling session.upload_progress for CVE-2026-109619. On PHP 8.5, PHP also recommends setting fatal_error_backtraces=0. Administrators should not use RLIMIT_AS or RLIMIT_DATA limits as a substitute, as these can trigger the vulnerable code path. These measures do not eliminate all exploitation paths. Updating to a fixed PHP release remains necessary. Until the next regular release, PHP also recommends keeping session.upload_progress disabled after patching to mitigate CVE-2026-115560, unless applications never change the session serializer at runtime. |
Why this matters
PHP runs a large share of the web, and much of it is not where security teams look first. It sits under content management systems and e-commerce platforms, inside container images, in hosting control panels, and in the web interfaces of appliances and network devices. Much of it is managed by a hosting provider, an operating system vendor or an appliance vendor rather than by the organization that depends on it.
Because the flaws are in PHP itself, they sit below the application. Patching WordPress, Magento or a custom application does nothing for this. The fix has to be applied to the PHP runtime, by whoever owns that runtime.
The timing puts defenders under immediate pressure. Public exploit code appeared before PHP released its fixes on 21 October, with the researcher reporting successful remote code execution and an ASLR bypass against a vulnerable Apache mod_php configuration. Attackers can now study and adapt that code without having to reverse-engineer the patch. Exploitation attempts are already widespread. The public exploits seen so far are narrow, each working only against a specific combination of operating system, PHP version and web server, but a more portable technique could appear at any time. Organizations should prioritize deploying the fixes immediately.
What organizations should do now
- Find every PHP runtime you are responsible for. Do not stop at web servers. Include container images, CI build images, serverless functions, development and staging hosts, and third-party products that bundle PHP. Record the exact version and how it was installed: an operating system package, an upstream build, a container image, a hosting provider or an appliance vendor. That determines who will ship your fix and when.
- Rank them by exposure. Prioritize Internet-facing PHP applications, especially those accepting file uploads, administration panels, or payment and customer data. Internal systems reachable from user networks come next.
- Identify what will not get a fix. Anything on PHP 8.1 or older is outside upstream support. Some Linux distributions and commercial providers backport security fixes to older branches, so check vendor advisories before assuming a system is unpatched based on its version number alone. Where there is no patch, restrict access to the system, isolate the system or migrate it to a supported branch. The official Docker php images for 8.1 and older will not receive this fix.
- Line up the people who patch for you. Confirm when hosting providers, managed service providers and appliance vendors will deliver updates. Operating system packages may lag behind upstream releases. Verify that the installed package includes the security fix, even if its version number differs from the upstream release.
- Prepare to deploy and verify the updates. Pre-approve emergency changes, prepare testing and build pipelines, and rebuild container images using updated PHP packages or base images. Verify the fixed runtime is actually deployed, and restart affected services or replace running containers to ensure they load the updated code.
- Reduce exposure while patches are pending. Remove unnecessary PHP applications from the internet and restrict administrative interfaces through VPNs or allow lists. Apply WAF and intrusion prevention protections, including targeted rules for these vulnerabilities where your vendor provides them, but do not assume generic rules will prevent exploitation. Limit request sizes and multipart nesting at the web server or proxy, and block multipart requests entirely for applications that do not accept uploads. Disabling session.upload_progress is not a confirmed workaround for both vulnerabilities and must not replace patching.
- Patch, then hunt. Begin monitoring exposed systems now and continue after updating. Look for unusual HTTP requests, unexpected PHP worker processes spawning shells or download tools, modified application files, web shells, and suspicious outbound connections. These are general indicators of compromise, not detections specific to these vulnerabilities. Prioritize investigation of systems that remained exposed before receiving the fix.
How TrendAI™ can help
- Shield vulnerable systems now with virtual patching. TrendAI Vision One™ TippingPoint™ provides Digital Vaccine filters. TrendAI Vision One™ Endpoint Security, including Server and Workload Protection, and TrendAI™ Deep Security, provide additional intrusion prevention. These rules shield vulnerable systems while updates roll out but do not replace patching.
- Detect exploit attempts on the network. TrendAI Vision One™ Network Detection and Response (NDR) detects PHP remote code execution exploit traffic.
- Decide where to act first. TrendAI Vision One™ Cyber Risk Exposure Management (CREM) maintains an attack surface view of the organization's devices, accounts, applications, cloud assets, and internet-facing assets. It assigns each asset a criticality level, and security teams can raise the criticality of the systems they know matter most. This gives a prioritized list of exposed systems to patch and investigate first.
- Look for signs of compromise. TrendAI Vision One™ raises Observed Attack Techniques and alerts for shell execution, web shells and reverse shells launched from PHP. Teams can also use XDR Data Explorer in TrendAI Vision One™ Agentic SIEM and XDR to search endpoint process activity for the behaviors described in "Patch, then hunt" above, such as PHP worker processes launching shells or download tools.
- See where you’re exposed. Cyber Risk Exposure Management flags assets running vulnerable PHP versions, and a Time-Critical Vulnerability alert in TrendAI Vision One™ notifies customers of affected assets.
- Protect containerized workloads. TrendAI Vision One™ Container Security includes runtime rules to detect exploitation activity in containers running PHP.
- Block follow-on activity. Behavioral monitoring and machine learning protections in Endpoint Security help prevent malicious payloads from running after exploitation.
- Hunt with ready-made queries. TrendAI Vision One™ Threat Intelligence includes an emerging threat entry for these vulnerabilities, with hunting queries teams can run in their own environment.
- Block known attack infrastructure. TrendAI™ Web Reputation Services blocks known attacker and scanner IP addresses and URLs associated with this activity.
- Hand it off. For teams without the capacity to run this themselves, TrendAI Vision One™ Services – Managed Detection and Response (MDR) provides 24/7 detection, investigation, and response by TrendAI™ analysts.
TrendAI™ customers with questions can contact their account representative. For others seeking guidance and help, please contact us at https://www.trendaisecurity.com/en-us/contact or email incident_response@trendmicro.com .