A significant Linux vulnerability was publicly disclosed on 29 April 2026. It is called Copy Fail, tracked as CVE-2026-31431, and it affects virtually every mainstream Linux server in production today. At the time of writing, no Linux distribution has shipped a patched kernel. A working exploit, 732 bytes of Python, is already public.
This is a news post. Things are moving quickly. We will update it as patches land.
What Happened
Security research firm Xint Code discovered a flaw in the Linux kernel that lets any ordinary user on a Linux machine promote themselves to full administrator, what Linux calls “root”. The attack works on Ubuntu, Red Hat Enterprise Linux, SUSE, Amazon Linux, Debian, Arch, Fedora, and other distributions running a kernel built since 2017.
The flaw was disclosed publicly on 29 April 2026. CERT-EU, the cybersecurity body for EU institutions, issued a high-severity advisory the same day. The CVSS severity score is 7.8 out of 10.
What This Means in Plain English
Most computer systems have two levels of user. There are normal users who can run programs and access their own files. Then there is the administrator account (called “root” on Linux), which can do anything: install software, read every file on the machine, change passwords, shut the system down, or silently exfiltrate data.
Copy Fail lets a normal user cross that boundary without permission in a matter of seconds. It does not require the attacker to know any password. It does not require physical access. It just requires that they can already run code on the machine.
That last point is important. This is what security professionals call a “local privilege escalation” vulnerability. It is not a way to break into a server from the internet. It assumes someone is already on the machine, either with a legitimate account or because a separate attack already gave them a foothold. Once they are in, Copy Fail takes them straight to the top.
Why It Works
The kernel is the core of the operating system. Every program on the machine, including your web server, database, and the SSH connection your team uses to log in, runs on top of it and trusts it to enforce boundaries.
The Linux kernel includes a cryptography interface called AF_ALG. It was designed to let normal programs use hardware-accelerated encryption without needing administrator access. In 2017, a performance optimisation was added to this interface. That optimisation has a logic bug: it allows data from the kernel’s own page cache, where the operating system stores files it is actively using, to be placed into a location that a normal user can write to.
By combining this with another kernel feature called splice(), a user can direct a small, controlled write at an arbitrary location in kernel memory. The exploit targets a setuid binary, typically /usr/bin/su, which is the program users run to switch accounts. By modifying a few bytes of that program in memory, the attacker causes it to hand them a root shell. The proof-of-concept published by the researchers does this in 732 bytes of Python and takes under a second.
The upstream kernel fix, which reverts the 2017 optimisation, was committed on 1 April 2026. However, that fix has not yet been packaged and shipped by any major Linux distribution.
Who Is Affected
Every Linux distribution running a kernel built between 2017 and the availability of a patched package is affected. The researchers directly verified the exploit on Ubuntu 24.04 LTS, Amazon Linux 2023, RHEL 10.1, and SUSE 16.
If you run any Linux server, whether self-hosted or in a cloud environment, on any of those distributions, you should treat it as affected until a patch arrives.
Android is not affected. Google’s Android enforces strict SELinux policies that prevent unprivileged processes from opening AF_ALG sockets at all. Ubuntu 26.04 (Resolute), released after the kernel fix, is also not affected.
Patch availability as of 30 April 2026: no fixed kernel package is available for Ubuntu 20.04 through 24.04, Amazon Linux 2023, SUSE Linux Enterprise, or Red Hat Enterprise Linux.
Is Your Business at Risk
If you do not run any Linux servers, you are not directly affected. This includes businesses that operate entirely on Windows, use only managed SaaS tools, and have no internal infrastructure beyond laptops and cloud-hosted software.
If you do run Linux, the key question is who can execute code on those machines. The attack requires local access, not remote access. That means you are most exposed if:
Your servers allow multiple users to log in with SSH. Any one of those users could escalate to root.
You run containers or Kubernetes clusters where untrusted workloads execute. A container escape combined with Copy Fail would give an attacker full control of the host.
You have CI/CD pipelines where code from external contributors runs on your infrastructure. A malicious pull request could exploit this before a human reviews it.
Your developers use shared Linux development environments. A single compromised developer account becomes a full system compromise.
If none of those apply to you, your exposure is lower. But the absence of a patch means the window of risk is open now.
What to Do Right Now
Apply the temporary mitigation. Before a patched kernel is available, you can disable the vulnerable kernel module entirely. This does not affect most applications. It does not break disk encryption (dm-crypt/LUKS), VPNs using IPsec, SSH, OpenSSL, GnuTLS, or TLS. The only applications it may affect are those explicitly configured to use the kernel’s AF_ALG interface directly, which is uncommon.
Run the following two commands on every affected Linux server:
echo "install algif_aead /bin/false" > /etc/modprobe.d/disable-algif.conf
rmmod algif_aead 2>/dev/null || true
The first command prevents the module from loading at boot. The second unloads it if it is currently running. To check whether any running process is using AF_ALG before you do this, run lsof | grep AF_ALG. On most servers this will return nothing.
If you use Docker, Podman, or Kubernetes, also add a seccomp policy that blocks AF_ALG socket creation. This provides a second layer of protection even on unpatched kernels, because the exploit requires opening an AF_ALG socket as its first step.
Check your cloud provider’s status page. AWS, GCP, and Azure are all working on patched kernel images. If you use managed compute services, your provider may be able to roll out a fix without you doing anything. Do not assume this has happened. Check.
Monitor for the patch. Sign up for security notifications from your distribution: Ubuntu: ubuntu.com/security/CVE-2026-31431 SUSE: suse.com/security/cve/CVE-2026-31431 Red Hat: access.redhat.com/security/cve/CVE-2026-31431
When a patched kernel arrives, reboot promptly. A kernel update only takes effect after a reboot.
Disclosure Timeline
- 2017: Vulnerable optimisation introduced into the Linux kernel (commit 72548b093ee3)
- 1 April 2026: Upstream kernel fix committed (commit a664bf3d603d)
- 29 April 2026: Xint Code publicly discloses the vulnerability and releases the proof-of-concept
- 29 April 2026: CERT-EU issues Security Advisory 2026-005
- 30 April 2026: No distribution has shipped a patched kernel
Conclusion
Copy Fail is a serious vulnerability and the absence of a patch from any major distribution means the risk window is open right now. The good news is the temporary mitigation is simple, low-risk, and takes under a minute to apply. If you run Linux servers, apply it today and watch your distribution’s security channel for the patched kernel.
If you are not sure whether your infrastructure is affected or need help assessing exposure across your environment, get in touch.
We can help you assess your exposure to Copy Fail and apply the right mitigations for your environment.
Get in touch →References
- Xint Code disclosure and technical writeup: xint.io
- CERT-EU Security Advisory 2026-005: cert.europa.eu
- Bugcrowd overview of CVE-2026-31431: bugcrowd.com
- The Hacker News coverage: thehackernews.com
- SUSE response: suse.com
- Proof-of-concept code: github.com/tgies/copy-fail-c